Search

US-12625254-B2 - Georadar imaging method and associated georadar

US12625254B2US 12625254 B2US12625254 B2US 12625254B2US-12625254-B2

Abstract

A method for imaging an area of interest on the ground uses a georadar equipped with a plurality of transmitter antennas and a plurality of receiver antennas. A bistatic RCS matrix is calculated at each point of a grid based on the matrix representing the MIMO channel modelling the propagation and the reflection in the area of interest, a matrix representing the losses along the propagation paths of the channel, and a phase-shifter matrix representing the delays on these same propagation paths. The bistatic RCS matrices relating to discrete frequencies belonging to the same coherence sub-band are summed and the elements of the matrices thus obtained are then summed incoherently to provide an overall backscatter coefficient for each point of the grid. Afterwards, an image representing this backscatter coefficient at each point of the grid is generated. A method for detecting a ground target-uses the same principle.

Inventors

  • Jean-Baptiste Dore
  • Raffaele D'ERRICO
  • Luc Maret

Assignees

  • Commissariat à l'énergie atomique et aux énergies alternatives

Dates

Publication Date
20260512
Application Date
20220623
Priority Date
20210624

Claims (10)

  1. 1 . A method for georadar imaging of an area of interest on the ground, the georadar operating in an analysis spectral band and being equipped with a plurality N of transmitter antennas as well as a plurality M of receiver antennas, the method comprising: meshing of the area of interest on the ground by a grid of points and decomposition of the analysis spectral band into a plurality Q of coherence sub-bands, each sub-band (B q ) comprising a set of discrete frequencies; calculation, for each point of the grid and each discrete frequency, of a matrix of losses A(p k , f, L) representing the attenuation of the signal transmitted by each transmitter antenna, having propagated to said point and then received by each receiver antenna; calculation, for each point of the grid and each discrete frequency, of a matrix of phasors, U(p k , f, L), each phasor corresponding to a propagation delay of the signal transmitted by each transmitter antenna, having propagated to said point and then received by each receiver antenna; estimation, for each discrete frequency, of the matrix, H(f), of the multiple in multiple out (MIMO) channel representing the N transmitter antennas, the area of interest on the ground and the M receiver antennas; estimation by channel equalisation of a bistatic radar cross section (RCS) matrix, {circumflex over (Γ)}(p k , f), for each point of the grid and each discrete frequency from the MIMO channel matrix for this frequency as well as the matrix of losses and the matrix of phasors, for this point of the grid and this frequency; coherent summation, over the discrete frequencies of each coherence sub-band, of the bistatic RCS matrices relating to a point of the grid, the coherent summation being performed for each coherence sub-band and each point of the grid so as to obtain a sub-band bistatic RCS matrix {circumflex over (Γ)}(p k , B q ) at each point of the grid; incoherent summation, over the different coherence sub-bands, of the elements of the sub-band RCS matrices, to obtain an overall backscatter coefficient at each point of the grid; generation of the image of the area of interest by representing the overall backscatter coefficient at each point of the grid.
  2. 2 . The georadar imaging method according to claim 1 , wherein the equalisation used to estimate the RCS matrix is a zero forcing (ZF) equalisation, an minimum mean square error (MMSE) type equalisation or an maximum combining ratio (MRC) type equalisation.
  3. 3 . The georadar imaging method according to claim 1 , wherein the signal transmitted in each coherence sub-band is an OFDM signal, the channel estimation in the discrete frequencies of this sub-band being carried out by means of pilot symbols modulating the sub-carriers of this signal.
  4. 4 . The georadar imaging method according to claim 1 , wherein the coherence bandwidth is selected smaller than or equal to the coherence bandwidth of the RCS of a predetermined target.
  5. 5 . The georadar imaging method according to claim 1 , wherein the matrix of losses A(p k , f, L) is calculated by A(p k , f, L)=G T (p k , f)⊙G R (p k , f)⊙B(p k , f, L) where G T (p k , f) is a gain of the transmitter antenna in the direction where it sees the point of the grid, G R (p k , f) is the gain of the receiver antenna in he direction where it sees the point of the grid and B(p k , f, L) models the losses in the medium, and ⊙ is Hadamard product.
  6. 6 . The georadar imaging method according to claim 1 , wherein said overall backscatter coefficient at the point of the grid p k is calculated by γ ^ ( p k ) = ∑ q = 1 Q ⁢ ∑ j = 1 M ⁢ ∑ i = 1 N ⁢ ❘ "\[LeftBracketingBar]" Γ ^ i , j ( p k , B q ) ❘ "\[RightBracketingBar]" 2 where {circumflex over (Γ)} i,j (p k , B q ) is the complex coefficient of the bistatic RCS for the coherence sub-band B q and Q is the number of coherence sub-bands in the analysis spectral band.
  7. 7 . A method for detecting a target on the ground by means of a georadar, said georadar operating in an analysis spectral band and being equipped with a plurality N of transmitter antennas as well as a plurality M of receiver antennas, further comprising: selection of a point of interest (p k ) on the ground and decomposition of the analysis spectral band into a plurality Q of coherence sub-bands, each sub-band (B q ) comprising a set of discrete frequencies; calculation, for each point of the grid and each discrete frequency, of a matrix of losses A(p k , f, L) representing the attenuation of the signal transmitted by each transmitter antenna, having propagated to said point and then received by each receiver antenna; calculation, for said point of interest and each discrete frequency, of a matrix of phasors, U(p k , f, L), each phasor corresponding to a propagation delay of the signal transmitted by each transmitter antenna, having propagated to said point and then received by each receiver antenna; estimation, for each discrete frequency, of the matrix, H(f), of the multiple in multiple out (MIMO) channel representing the N transmitter antennas, the point of interest on the ground and the M receiver antennas; estimation by channel equalisation of a bistatic radar cross section (RCS) matrix, {circumflex over (Γ)}(p k , f), for the point of interest and each discrete frequency from the MIMO channel matrix for this frequency as well as the matrix of losses and the matrix of phasors, for this point of interest and this frequency; coherent summation, over the discrete frequencies of each coherence sub-band, of the bistatic RCS matrices, the coherent summation being performed for each coherence sub-band so as to obtain a sub-band bistatic RCS matrix {circumflex over (Γ)}(p k , B q ) for said point of interest; incoherent summation, over the different coherence sub-bands, of the elements of the sub-band RCS matrices, to obtain an overall backscatter coefficient at said point of interest; comparison of said overall backscatter coefficient with a predetermined threshold value, a target being detected at the point of interest if the overall backscatter coefficient is higher than the threshold value, and not being detected otherwise.
  8. 8 . The method for detecting a ground target by means of a georadar according to claim 7 , wherein the equalisation used to estimate the RCS matrix is a zero forcing (ZF) type equalisation, an minimum mean square error (MMSE) type equalisation or an maximum combining ratio (MRC) type equalisation.
  9. 9 . The method for detecting a ground target by means of a georadar according to claim 7 , wherein the signal transmitted in each coherence sub-band is an orthogonal frequency-division multiplexing (OFDM) signal, the channel estimation in the discrete frequencies of this sub-band being carried out by means of pilot symbols modulating the sub-carriers of this signal.
  10. 10 . The method for detecting a ground target by means of a georadar according to claim 7 , wherein said overall backscatter coefficient at the point of the grid p k is calculated by γ ^ ( p k ) = ∑ q = 1 Q ⁢ ∑ j = 1 M ⁢ ∑ i = 1 N ⁢ ❘ "\[LeftBracketingBar]" Γ ^ i , j ( p k , B q ) ❘ "\[RightBracketingBar]" 2 where {circumflex over (Γ)}(p k , B q ) is the complex coefficient of the bistatic RCS for the coherence sub-band B q and Q is the number of coherence sub-bands in the analysis spectral band.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This is a National Stage application of PCT international application PCT/FR2022/051242, filed on Jun. 23, 2022, which claims the priority of French Patent Application No. 2106797, filed Jun. 24, 2021, both of which are incorporated herein by reference in their entirety. TECHNICAL FIELD The present invention relates to the field of ground penetration radars also called geological radars or georadars, and more particularly that of ground imaging by such radars. PRIOR ART Georadars are commonly used in demining and detection of buried networks. The evolution of regulations in some European countries aims to reinforce the obligation to detect the buried networks when requesting authorisation for works or to map sensitive networks in urban areas. Different methods for georadar ground imaging are known from the prior art. In general, they consist in transmitting a radar signal into the ground by means of a transmitter antenna in several successive positions, or from a plurality of transmitter antennas, and in receiving the reflected signal by means of an antenna in several successive positions or several receiver antennas. Afterwards, the received signals are subjected to a so-called focusing or migration processing intended to perform position or phase shifts on the signals reflected by different objects in the ground. For example, the migration techniques used in this processing include Kirchhoff migration and F-k migration a description of which will be found in the article by N Smitha et al. entitled “K Kirchhoff and F-k migration to focus penetrating radar images” published in Int'l Journal of Geo-Engineering, Vol. 7, No. 4, April. 2016. Nonetheless, these migration imaging methods have some drawbacks. First of all, these are sensitive to couplings whether to couplings between the antennas or to couplings between the antennas and the ground. Besides, they do not allow easily resolving several layers of objects in the ground. For example, if networks are buried at different depths, that one buried deeply will be masked by that one which is close to the surface of the ground. Finally, they require a uniform meshing of the area of interest, which is sometimes useless. A first object of the present invention is to provide a ground imaging method by georadar which does not have the aforementioned drawbacks and which in particular allows avoiding the coupling effects as well as solving layers of buried objects at different depths. A second object of the present invention is to detect the presence and to locate objects present in the ground with a lower false alarm rate than in the prior art. DISCLOSURE OF THE INVENTION The present invention is defined by a method for georadar imaging of an area of interest on the ground, said georadar operating in an analysis spectral band and being equipped with a plurality N of transmitter antennas as well as a plurality M of receiver antennas, said imaging method comprising: meshing of said area of interest on the ground by a grid of points and decomposition of the analysis spectral band into a plurality Q of coherence sub-bands, each sub-band (Bq) comprising a set of discrete frequencies;calculation, for each point of the grid and each discrete frequency, of a matrix of losses A(pk, f, L) representing the attenuation of the signal transmitted by each transmitter antenna, having propagated to said point and then received by each receiver antenna;calculation, for each point of the grid and each discrete frequency, of a matrix of phasors, U(pk, f, L), each phasor corresponding to a propagation delay of the signal transmitted by each transmitter antenna, having propagated to said point and then received by each receiver antenna;estimation, for each discrete frequency, of the matrix, H(f), of the multiple in multiple out (MIMO) channel representing the N transmitter antennas, the area of interest on the ground and the M receiver antennas;estimation by channel equalisation of a bistatic radar cross section (RCS) matrix, {circumflex over (Γ)}(pk, f), for each point of the grid and each discrete frequency from the MIMO channel matrix for this frequency as well as the matrix of losses and the matrix of phasors, for this point of the grid and this frequency;coherent summation, over the discrete frequencies of each coherence sub-band, of the bistatic RCS matrices relating to a point of the grid, the coherent summation being performed for each coherence sub-band and each point of the grid so as to obtain a sub-band bistatic RCS matrix {circumflex over (Γ)}(pk, Bq) at each point of the grid;incoherent summation, over the different coherence sub-bands, of the elements of the sub-band RCS matrices, to obtain an overall backscatter coefficient at each point of the grid;generation of the image of the area of interest by representing the overall backscatter coefficient at each point of the grid. The equalisation used to estimate the RCS matrix could be a