EP1119779A1

EP1119779A1 - System for detection of objects in the ground

Info

Publication number: EP1119779A1
Application number: EP99949200A
Authority: EP
Inventors: Peter Donald Fraser Tait; David John Gunton; Timothy Edward Ffrench
Original assignee: Marconi Electronic Systems Ltd
Current assignee: BAE Systems Electronics Ltd
Priority date: 1998-10-12
Filing date: 1999-10-11
Publication date: 2001-08-01
Also published as: WO2000022455A1; AU6217899A

Abstract

A detection system (1) comprises a radar unit (2) connected to an array controlling network (3) which is in turn connected to an array (4) of sensors (5). The array (4) consists of a number of sensors (5) each comprising a transmit-receive antenna pair. The radar unit (2) consists of a signal source, a receiver and an analogue-to-digital converter. The signal source generates a high resolution waveform such as a broadband chirp which is transmitted to each of a number of ground positions by the array (4). A return radar signal is reflected from the ground to the array (4) for processing in order to determine the location of an object, for example a mine, located in, on or at the surface of the ground. The location of an object is determined by measuring the spatial correlation of the returned radar signal at adjacent positions along the area of ground to be investigated.

Description

SYSTEM FOR DETECTION OF OBJECTS IN THE GROUND

This invention relates to the detection of objects located at, on or adjacent to the surface of the ground and is particularly related to the detection of objects such as mines buried in the ground.

In the past mines typically contained a substantial proportion of metal and could be detected by conventional metal detectors. However, modern mines comprise a substantial proportion of plastics material and little or no metal. What metal there is is often insufficient to be readily detected by metal detectors. Although increasing the sensitivity of the metal detector increases the chances of detecting modern mines, it also increases the chances of detecting non-threatening metal objects such as small pieces of metal scrap.

Ground penetrating radar systems are known which use an impulse waveform to detect mines. However such systems have a relatively low frequency spectral content and so the wavelength at the spectrum peak is typically large compared with the dimensions of the mine. As a result, the mine exhibits a low radar cross-section. Although such systems can detect large mines, for example anti-tank mines, they have difficulty in detecting smaller mines such as anti-personnel (AP) mines. Anti-personnel mines typically have a diameter of about 100mm.

According to the invention there is provided an object detection system for detecting the presence of objects in, on or at the surface of ground, comprising a transmitter of a radar signal, a receiver of a returned radar signal and processing means characterised in that the processing means is arranged to detect objects by measuring the spatial correlation of the received signal at adjacent points of the ground.

Preferably, the object is a mine, and most preferably an anti-personnel mine.

The frequency of the transmitted radar signal may be arranged to vary during the time of its emission. The transmitted radar signal may be in the form of a chirp.

The system may comprise an array of transmitters and receivers. The array of transmitter and receivers may form a phased array.

The system may be carried for use by a user. The array of transmitters and receivers are arranged be hand held by a user.

An embodiment of the invention will now be described by way of example only with reference to the accompanying drawings in which:

Figure 1 schematically illustrates a detection system according to the invention; Figure 2 illustrates an antenna array for a detection system, and

Figure 3 illustrates a unit for generating and receiving a radar signal.

A detection system according to the invention is shown in Figure 1. The detection system 1 comprises a radar unit 2 connected to an array controlling network 3 which is in turn connected to an array 4 of sensors 5. The array 4 consists of a number of sensors 5 each comprising a transmit-receive antenna pair. The radar unit 2 consists of a signal source (including a trigger), a receiver and an analogue-to-digital converter. The signal source generates a high-resolution waveform such as a broad-band chirp. An advantage of using a chirp waveform is that it can be characterised and its spectrum controlled relatively easily in the frequency range of particular interest, for example greater than 1GHz. It is more difficult to characterise and control a short impulse-like waveform at this kind of frequency. The detection system transmits a radar signal towards the ground. A signal is returned from the ground and may include a clutter elements, that is an element of the signal is reflected signal from the ground's surface, scattered elements, from within the ground itself, and object elements scattered from mines. The received signal is processed in a signal processor 6 which provides the results of a detection operation, for example visually.

The sensor array 4 may be in the form of a single row of antennas, more than one row of antennas or antennas arranged in a cluster, for example having groups in which a single antenna is surrounded by six nearest neighbours having centres in a hexagonal arrangement. Figure 2 shows an arrangement having a pair of rows.

The nature of the array controlling network 3 depends on how the array is to be operated.

In an embodiment in which the sensors 5 are arranged to operate independently, the array controlling network 3 will be a switching arrangement. The switching arrangement switches individual, or specific groups of, sensors 5 to operate at one time and then other individual, or specific groups of, sensors 5 to operate at other times. In this way over a time period in which all of the sensors 5 have been operated, a detection operation will have occurred, over the array 4. In another embodiment the transmit antennas can be arranged to work in concert so that a beam or beams produced by the array 4 can be steered. Alternatively only a single wide angle transmit antenna, or several wide angle transmit antennas, may illuminate an area of ground to be examined and a greater number of receive antennas may be operated co-operatively so as to steer the direction of maximum sensitivity around the area of ground to be examined. Electronic steering of arrays is well known in the art and requires an array controlling network having switching and phase shift or time delay elements. For wideband waveforms, time delay elements are preferable to phase shift elements and may be provided by different cable lengths or, for example, by a chirped Bragg fibre grating with an associated tunable laser. Hence, the time delay between individual sensors or groups of sensors may be varied continuously or in steps, to steer or scan the radar beam. Although the array 4 of Figure

1 shows a single radar unit 2, a number of such units may be provided each driving one or more sensors 5.

A suitable radar unit 2 is described in patent application GB 2317769. Such a radar unit

2 is described below in relation to Figure 3. It comprises a transmitter 8 and a receiver 9. In the transmitter 8, a digital waveform generator 10 produces cosine 11 and sine versions 13 of a chirp waveform of limited duration. This waveform is repeated at a preset frequency. Each of the cosine and sine versions are of 500MHz bandwidth. They are fed to the in-phase (I) and quadrature (Q) ports of a vector modulator 14 driven from a 4GHz local oscillator 16. The vector modulator 14 produces a waveform output which is a chirp pulse with a 1GHz bandwidth centred at 4GHz, that is, the frequency within the pulse varies from 3.5GHz to 4.5GHz. The waveform output is passed through a bandpass filter 18 to remove any undesired out-of-band spurious signals. As a result, a waveform having an intermediate frequency is produced which can be processed further.

The waveform is amplified by an amplifier 20. A coupler 22 divides the waveform into two signals which are used to drive an intermediate frequency (I.F.) port 24 and a local oscillator (L.O.) port 26 of a wideband mixer 28. The coupler 22 has a coupling value which is chosen so that the signal at the L.O. port 26 is a high level signal with power in the range specified for that port 26, and the signal at the I.F. port 24 is a low level signal, for example 15 to 20dB lower in power than the signal used to drive the L.O. port 26.

The two signals take respective paths 25, 27 from the coupler 22 to the ports 24, 26 of the mixer 28. The time delay and phase length of the two paths 25, 27 should be matched so that an output signal is produced which has the same time duration as the two input signals and the centre frequency and within-pulse modulation bandwidth are twice the input frequency and bandwidth. The output signal now has a within-pulse chirp of 7 to

9GHz. It may be necessary to include an adjustable line length in one or both of the paths 25, 27 from the coupler 22 to the mixer 28.

The coupler 22 and mixer 28 in combination act as a multiplier which produces an output signal having (in this embodiment) a bandwidth and frequency twice that of the waveform which was originally available at the output of the vector modulator 14.

The output signal passes through a filter 30 and is filtered to remove any undesired out- of-band spurious signals generated by the mixer 28. It is then amplified by an amplifier 32 to a level suitable for a further multiplication step. This is performed by a second combination of a coupler 34 and mixer 36 which operates similarly to the combination described above. Again the frequency and bandwidth are both multiplied by a factor of two.

The second combination of coupler and mixer (serving as a second bandwidth multiplier) produces an output signal which has a within-pulse chirp of 14 to 18GHz. The output signal from the second mixer 36 is filtered and amplified by a filter 38 and an amplifier 40. The output signal is passed through a coupler 42 and a sample of the output signal is coupled off to be used in the receiver 9. This will be discussed below. A portion of the output signal which remains after coupler 42 is fed to a LO port 44 of a wideband mixer 46. An IF port 47 of the mixer 46 is driven from a CW oscillator 48 to produce an output signal at a RF port 50. The within-pulse chirp modulation will be centred on a frequency which depends upon the frequency of the oscillator 48. This output signal is passed through a bandpass filter 52 to select the upper or lower sideband and to remove spurious signals, and is transmitted as a radar signal from a transmitting horn antenna 54. Typically, the oscillator 48 can have a frequency of 6GHz and the lower sideband is selected so that the transmitted radar signal has a chirp pulse varying from 8 to 12GHz.

Any returned signal is received by the receiver 9. This has a receiving horn antenna 56 which feeds any returned signal to an RF port 58 of a wideband mixer 60. The sample of the output signal taken from coupler 42 is passed through a variable delay 62 which serves as a range gate control and is then fed to a LO port 59 of the wideband mixer 60. If required, suitable gating can be included in order for the system to sweep through the range swath of interest by varying the delay. The output from an IF port 61 of the mixer 60 is then filtered by a bandpass filter 64 centred at 8GHz, before being passed through a low noise amplifier 66. The amplifier 66 together with the losses of the components which precede it determines the noise figure of the receiver 9. If a more sensitive receiver 9 is required a low noise amplifier should be included between the receiving horn antenna 56 and the wideband mixer 60.

Down-conversion to baseband is performed in a quadrature mixer configuration comprising a zero degree splitter 68, a 90° coupler 70 and two mixers 72, 74. The mixers 72, 74 are matched in gain and phase to produce I and Q signals at baseband. The baseband signals are integrated in low pass filters 76, 78 which are matched to the chirp pulse length and amplified by amplifiers 80, 82 to a level suitable for the analogue-to- digital convertors 84, 86 which digitise their respective received signals 81, 83. Digitised signals are stored in computer memory 88 for further off-line signal processing which may include Doppler processing.

The receiver 9 acts as a correlator with a multiplication process performed in the wideband mixer 60 at the RF stage and an integration process performed in the low pass filters 76, 78 at baseband. The power level at the output of the correlation receiver 9, represented by the sum of the squares of the instantaneous amplitudes of signals 81 and

83 is at a maximum value if the variable delay which is added to the sample of the output signal corresponds to the round trip range delay to the object. At all other object ranges, that is those for which the variable delay does not correspond to the round trip range delay, the output signal level is low, given by the off peak level of the waveform auto- correlation function. A well designed waveform with 4GHz bandwidth will give a peak with 3dB width corresponding to a range resolution of about 50mm and sidelobe levels at least 40dB down from the peak.

If the range gate width is 50mm in air the range gate width in the ground will be reduced because the signal propagation velocity in the ground is reduced by the square root of the ground's dielectric constant. For example, for a dielectric constant of 8, the range gate width in the ground will be approximately 18mm. Using such a resolution should enable undesirable surface reflection elements in the return signal to be isolated from desirable sub-surface reflection elements in the return signal, for example mines, by analysing separate range gates. In any case a high return element will be reflected from the ground's surface which will, to some degree, mask the signal element scattered from objects in the ground. To overcome this problem the antennas of each transmitter- receiver pair may be mounted with boresight orientated 30° to the vertical to give an angle of 60° incidence to the ground's surface, so that the reflected element of the signal from the ground surface is not returned to the receive antenna.

The system 1 can transmit a frequency and a waveform which are adaptively tuned for the detection of specific objects in specific environments. Different radar waveforms can be programmed for optimum detection of particular objects in particular conditions.

Mode scheduling can be used to transmit sequences of different waveforms for detecting different types of objects almost simultaneously.

The system 1 is coherent and so is compatible with Synthetic Aperture Radar and Doppler beam sharpening methods. This enables real time maps of the ground to be produced as the array 4 is moved forward.

A benefit of using the system 1 described above is that it enables a signal having a wide instantaneous bandwidth, such as a 3 or 4GHz chiφ, to be put on a microwave frequency carrier. For landmine detection this carrier frequency may be in the range 8 to 18GHz. In order to compromise between resolution and attenuation a frequency of 10GHz may be preferred. A relatively high carrier frequency provides several benefits. The high (microwave) frequency which is used has a wavelength which is of the same order as AP mine dimensions. As a consequence the radar cross-section of the mine is in the resonance detection region and so detection sensitivity is maximised. The high frequency allows the use of physically small antennas (for example TEM horns). The aperture dimension of the antennas can be about 50mm rather than about 300mm which is typically used in impulsive systems. These small antennas provide narrow radar beams, for example 20° at the 3dB points. They also provide a relatively small radar footprint on the ground comparable to the object size to aid clutter rejection. The small ground footprint gives good resolution in two orthogonal horizontal dimensions giving a resolution cell matched to the object dimensions, and allowing a tight grid pattern to enable spatial coherence effects to be used as a discrimin.ant. Furthermore, such narrow, low divergence, beams allow for a stand-off distance (for example 400 to 500mm) from the ground which is large enough to avoid ground impediments. An advantage of this is that it enables phased arrays to be used which typically have to be several wavelengths, for example ten, away from the object. A cluster of antennas forming an array 4 can be small enough to be handheld. A further advantage is that the system 1 has a lower fractional bandwidth than known systems. Therefore the components required in the system .are easier to design than in known systems.

Although such high frequencies provide reduced ground penetration due to the higher attenuation in the ground, this is not a major problem for the detection of small antipersonnel mines either at the surface or buried at shallow depths. Such mines are normally placed at a shallow depth since they are ineffective if buried too deep.

If the ground is uniform, the signal returned from it (in the absence of any buried object) will not change greatly as an individual antenna, or an array of antennas, makes measurements across it. The presence of an object in such ground will result in a clear indication of that object. However, in normal circumstances the ground will not be uniform and differences will emerge as a series of measurements are made. At small distances between measurements they will not differ much. As the distance is increased the difference also increases until eventually a distance is reached at which the measurements are essentially independent. The technique used takes a series of measurements and looks for correlation between them. If the measurements relate to adjacent, or overlapping, areas of ground not containing an object, there will be little correlation and the system 1 will indicate apparently clear ground. If an area of ground is measured in which an object is located, a series of measurements encompassing that object will have some, that is a relatively high, degree of correlation as compared to statistically insignificant correlation. Effectively, the system 1 looks for parts of the ground which are similar to each other. As a consequence the system 1 will indicate when a position in the ground contains an object which is more uniform in its dielectric properties than the ground around that position.

The correlation distance depends on the nature of the ground and the footprint. If the ground is of a rapidly varying character then the correlation distance will be small, that is, if a particular area of ground is being considered, only a small movement of the antenna array 4 will cause it to be examining ground which is very different. In these circumstances, the correlation distance will be approximately equal to the diameter of the footprint. A small footprint, of a size less than or approximately equal to the size of the objects being sought, is advantageous because an object located within it occupies a relatively greater proportion of it and thus provides greater correlation between adjacent readings. Since it is difficult to obtain smaller and smaller footprints it should be understood that the system 1 works effectively when the footprint is broadly the same size as the object although a smaller footprint could provide greater sensitivity.

Since the system 1 is arranged to identify different areas of ground having a high degree of correlation, it is necessary for it to determine when such correlation is caused by the ground being relatively uniform. Image processing algorithms use information relating to the dimensions of objects being sought and work out when a uniform signal over a greater distance suggests that an object is not present. In such an environment, a buried object will be easier to spot in ground which is relatively non-uniform and so to detect an object the system looks for its edge. An edge of an object can be indicated by a localised reduction in the correlation coefficient.

An advantage of the system 1 using a phased array technique is that it provides a small footprint because the aperture dimension of the 4 array is large. Another advantage of such a technique is that the ground can be scanned electronically without the need to move the array 4 carefully over the ground. This reduces the risk of a manual operator missing part of an area of ground to be examined.

For a hand held system, a cluster of horn antennas may be integrated (perhaps with other sensor types) allowing array processing with non-critical handling because the area covered by electronic steering is considerably greater than the dimensions of the footprint.

However, if a phased array technique is not used, and the array 4 comprises a number of separately operating sensors 5, in order to scan an area of ground the array 4 has to be moved physically over the ground. In such a system 1 it is desirable to collect data at a grid spacing approximating to the length of the correlation distance, for example 50 to 100mm. In the array shown in Figure 2, the aperture dimension of each antenna is

100mm. Therefore, a staggered arrangement of sensors 5 or antennas is used in which adjacent rows are displaced from each other by 50mm in a lateral direction across the array 4, that is parallel to the rows. As the array is moved forwards, a detection operation is carried out at 50mm intervals. It should be understood that the number of rows required in the array 4 depends on the aperture dimension of the antennas. If an aperture dimension of more than 100mm is used, more rows having a smaller degree of offset from their neighbours will be required in order for the array 4 to provide sufficient closeness of measuring points in the lateral direction so as to provide a measurement dimension corresponding to the correlation distance. In this event, to maintain reliability of measurement by the system 1, the array 4 would be moved forward a correspondingly smaller distance (equivalent to the offset) at each set of measurements. This system 1, requiring controlled movement, is especially suitable for mounting on a vehicle. This allows mounting of a large array 4. If the array 4 is statically mounted on a vehicle then the staggered arrangement of Figure 2 is necessary. Alternatively, if the array 4 can be stepped rapidly in a lateral direction through a number of positions as the vehicle moves forward slowly, the number of rows can be reduced, perhaps to a single row.

The received signal is processed in real time to identify objects based on their size and shape. A visual display presents this information to an operator for identification of objects. In one embodiment the processor 6 automatically examines objects and triggers an automatic alarm on locating a potential threat, such as a mine

In one embodiment the array 4 is separated into a number of sections, for example five, to allow the array 4 to follow the terrain by independently controlling the vertical position of each section. The need to separate the array 4 in sections will depend on the tolerance of the antenna design to ground clearance.

Claims

1. An object detection system for detecting the presence of objects in, on or at the surface of ground, comprising a transmitter of a radar signal, a receiver of a returned radar signal and processing means characterised in that the processing means is arranged to detect objects by measuring the spatial correlation of the received signal at adjacent points of the ground.

2. An object detector system, as in Claim 1, wherein the object is a mine.

3. An object detector system, as in Claims 1 or 2, wherein the object is an antipersonnel mine.

4. An object detector system, as in any preceding claim, wherein the frequency of the transmitted radar signal is arranged to vary during the time of its emission.

5. An object detector system, as in Claim 4, wherein the transmitted radar signal is in the form of a chiφ.

6. An object detector system, as in any preceding claim, wherein the system comprises an array of transmitters and receivers.

7. An object detector system, as in Claim 6, wherein the array of transmitter and receivers form a phased array.

8. An object detector system, as in any preceding claim, wherein the system can be carried for use by a user.

9. An object detector system, as in Claims 6 or 7, wherein the array of transmitters and receivers are arranged be hand held by a user.

10. An object detector system substantially as illustrated and/or described with reference to the accompanying drawings.