GB2458465A

GB2458465A - Remote detection of one or more dimensions of a metallic or dielectric object

Info

Publication number: GB2458465A
Application number: GB0804944A
Authority: GB
Inventors: Nicholas Bowring; David Andrews; Nacer Ddine Rezqui; Stuart Harmer
Original assignee: Manchester Metropolitan University
Current assignee: Manchester Metropolitan University
Priority date: 2008-03-18
Filing date: 2008-03-18
Publication date: 2009-09-23
Anticipated expiration: 2028-03-18
Also published as: GB2458465B; GB0804944D0; GB2458465A9; GB2458465B8

Abstract

A portable microwave or millimetre wave arrangement is provided for remotely detecting one or more dimensions of a metallic or dielectric object. The object may include a weapon or an explosive concealed on or in the body of a person, within a container or in an item of luggage. Microwave or mm wave radiation is directed towards a target and radiation is received from the target and detected in the frequency domain. The directed radiation is swept over a predetermined range of frequencies and the detected signal is transformed into the time domain such that one or more dimensions of a detected object can be determined. In a preferred embodiment a wide frequency scan, signal processing techniques and polarisation effects are used to enhance the detection process in a field deployable system.

Description

Remote detection and measurement of objects

Technical field

The present invention relates to the detection of objects, and more particularly to techniques for remote detection and measurement of objects.

The invention has various applications in scenarios where metallic objects are to be detected, but is particularly useful in security equipment, for example at airports and other public areas.

Background art

It is well known to use electromagnetic radiation to detect the presence of objects (e.g. handheld detectors used for detecting objects on or under the ground, and walk-through arches at airports).

However, the conventional detectors used at airports are unable to determine the dimensions of objects to any significant degree, and are thus unable to distinguish between objects of different types, i.e. harmless (belt buckles, cameras), and potentially dangerous (guns, knives).

The detection of concealed weapons, especially handguns, is a very great problem for security applications that currently cannot be policed without a non-portable system, for example random checks in an urban environment. The use of microwaves (electromagnetic * ** waves with wavelengths in the centimeter to millimeter range) provides a means for the * * * -. standoff detection and identification of concealed conducting items such as handguns and *SS.

knives. Large metal objects, such as hand guns, give a significantly different and generally * larger response when irradiated by low power microwaves than thaL from the human body, * .. clothing and benign normally-carried objects, and this can be detected using a combination of antenna and sensitive receiver. S..

: By actively illuminating an object with wide-range swept or stepped frequency microwave or mithmetre wave radiation, the frequency response of the return signal gives the range and also information about the dimensions of the object, This method is equivalent to using a fast microwave pulse and measuring the response as function of time, as used in conventional RADAR. Selecting a part of the return signal within a particular range aids the positive identification of the suspect object and also helps to reject background signals. The analysis of the time response gives further information as to the dimensions of the target. This technique can also be applied to the detection of dielectric layers, such as an explosive vest strapped to a suicide bomber (see Active millimeter wave detection of concealed layers of dielectnc material, Bowling N. J., Baker J. G., Rezgui N., Southgate M., Proceedings of the SPIE 6540-52 2007; and A sensor for the detection and measurement of thin dielectric layers using reflection of frequency scanned millimetric waves, Bowring N.J., Baker J.G., Rezgui N., Alder J.F. Meas. Sci. TechnoL 19 024004 (7pp) 2008). However, such techniques have not been heretofore used for detecting and measuring metal objects.

A practical system based on swept frequency RADAR has been proposed (US Patents 6.359,582 and 6,856,271 and published application US200710052576). In the disclosed systems, the frequency is swept by typically by I GHz around about 6 GHz. The depth resolution that is achievable is therefore only 15 cm, thus the system cannot give details of the objects. The detection relies on companng gross features of the signal as a whole with similar suspicious and benign signals to which the system had been previously exposed.

Also disclosed is the measurement of polarization properties of the scattered signal.

In the aforementioned patent documents, the low frequency of operation makes the angular resolution of the antennae poor and the wide field of view makes It difficult to single out particular targets or to determine on which part of the target the threat is situated. This may be improved by changing to higher frequencies where microwave optics becomes effective.

This is particularly important for explosives detection where the contrast from the body signal is low. Systems working at higher frequencies but still with a limited bandwidth have been proposed by Gorman et at (US Patent 6,967.612) and by Millitech (US Patent 5,227,800).

Many systems have been produced to enable images of the target to be obtained using either active microwave illumination or the passive thermal emission of the target (SPIE 2007).

These systems use multi-detector arrays and some form of mechanical scanning. Passive * ** systems, though giving more realistic images, tend to be slow and show poor contrast for * * * dielectric targets. Active illumination systems can be acquired faster, but suffer from strong ***.

** * S reflections from benign objects such as the human body, which make it difficult to distinguish from metal threat objects. AU scanning systems require compiex human or Artificial * * Intelligence interaction to interpret the image and to pick out the suspect features. This makes their deployment in many applications difficult. As used herein, "threat object" is taken to mean a metallic or dielectric object, whether specifically designed or intended for offensive **5 * : use or not, that have potential to be used in an offensive or violent manner. * P...

It is apparent that systems which can identify threat objects at stand-off have many applications, where conventional metal detector booths are inappropriate. These indude covert surveillance and mobile operation in streets and buildings. Portable, compact and cost-effective systems are not presently available and this invention seeks to address this need.

Summary of the invention

According to one aspect of the present invention there is provided a system for remote detection of one or more dimensions of a metallic or dielectric object, comprising: a transmission apparatus, including a transmission element, for directing microwave or mm wave radiation in a predetermined direction, a detection apparatus, for receiving radiation from an entity resulting from the transmitted radiation and generating one or more detection signals in the frequency domain, and a controller, the controller being operable to (i) cause the transmitted radiation to be swept over a predetermined range of frequencies, (ii) perform a transform operation on the detection signal(s) to generate one or more transformed signals in the time domain or optical depth domain, and (iii) determine, from one or more features of the transformed signal, one or more dimensions of a metallic or dielectric object upon which the transmitted radiation is incident.

Preferably, the transmission element is a directional element that may be pointed by a user.

Preferably, the controller is operable to initiate step (i) upon receiving an activation signal, the activation signal corresponding to a user input or to detection of the presence of the entity.

Preferably, step (i) comprises stepwise sweeping by predetermined steps in frequency; and step (ii) comprises (iia) perform a transform operation after each sweep to produce a time domain or optical depth domain trace, and (iib) storing each time domain trace in a respective sweep channel, said time domain or optical depth domain traces thereby comprising said transformed signals.

Preferably, step (iii) comprises normalising said according to range one or more transformed signals.

Preferably, step (iii) further comprises using a Complex Fourier Transform or Direct Fourier Transform to convert transformed signals to the x-dimension, and determining the x-positions of peaks on the transformed signals. * S.

Preferably, step (iii) further comprises, from the x-positions, using S.. _c

W

where L=distancetoentity c = the speed of light tf= the periodicity in the frequency domain, to determine corrected x-axis positions, and thereby optical depth.

In one embodiment, step (ii) includes the procedure of Appendix B. 1, thereby producing first and second outputs (Outputl, Output2) dependent upon the detection signals, wherein Outputl is the sum of all correlations between vectors in a first array, the vectors in the first array comprising, for each sweep channel, a stored signal above a threshold that are derived by Direct Fourier Transform from the detection signals, and Output2 is the sum of integrated signals above the threshold for each sweep channel.

In another embodiment, step (ii) includes the procedure of Appendix B.2, thereby producing a third output (Output3) dependent upon the detection signals, wherein Output3 is, for each sweep channel, the best (lowest) correlation value between the ideal response for a number of barrel lengths and for a number of weapon calibres stored in memory and the direct untransformed detection signals IERI2.

In another embodiment, the detection apparatus includes a first detection element directed in a first direction, towards the entity, for generating non-polansed detection signals, and a second detection element, directed at 90 degrees to the first element, for generating cross-polarised detection signals, and wherein step (ii) includes the procedure of Appendix B.3, thereby producing fourth and fifth outputs (Output4, Output5) dependent upon the detection signals, wherein Output4 is, for each sweep channel, the sum of correlations between the transformed signals, the transformed signals comprising a Complex Fourier Transform of the non-polarised detection signals and of the cross-polarised detection signals, and Output5, for each sweep channel, the sum of the Integrated non-polarised and cross-polansed signals after Complex Fourier Transform. Preferably, the integrated non-polarised and cross-polansed signals comprise integrations of transformed signals within one or more distance windows, the contents of each distance window being stored in a database in association with a respective response for a particular weapon.

In another embodiment, step (ii) includes the procedure of Appendix B.4, thereby producing sixth and seventh outputs (Output6, Output7) dependent upon the detection signals.

Preferably, the system further includes a neural network, the neural network having as inputs thereto any combination of (a) the first and second outputs (Outputl, Output2), (b) the third : output (Output3), (c) the fourth and fifth outputs (Output4, Output5), and (d) the sixth and seventh outputs (Output6, Output7), wherein the output of the neural network is an indication *. : of a confidence level of a metallic or dielectric object of a predetermined type being detected, * : ** for example, I =gun detected, 0= no metallic or dielectric object detected.

According to another aspect of the present invention there is provided a method for remote detection of one or more dimensions of a metallic or dielectric object, comprising: a -4.

transmission apparatus, including a transmission element, for directing microwave or mm wave radiation in a predetermined direction, providing a detection apparatus, for receiving radiation from an entity resulting from the transmitted radiation and generating one or more detection signals in the frequency domain, and providing a controller, and operating the controller to (i) cause the transmitted radiation to be swept over a predetermined range of frequencies, (ii) perform a transform operation on the detection signal(s) to generate one or more transformed signals in the time domain, and (iii) determine, from one or more features of the transformed signal, one or more dimensions of a metallic or dielectric object upon which the transmitted radiation is incident.

Preferably, step (i) comprises stepwise sweeping by predetermined steps in frequency; and step (ii) comprises (ha) perform a transform operation after each sweep to produce a time domain or depth domain trace, and (iib) storing each time domain or depth domain trace in a respective sweep channel, said time domain or depth domain traces thereby comprising said transformed signals.

Preferably, step (iii) comprises normalising according to range one or more transformed signals.

Preferably, step (iii) further comprises using a Complex Fourier Transform or Direct Fourier Transform to convert transformed signals to the x-dimension, and determining the x-positions of peaks on the transformed signals.

Preferably, step (iii) further comprises, from the x-positions, using

C 24f p. *. I* *

* where L = distance to entity c= thespeedof light f= the periodicity in the frequency domain, *.*S *..* : to determine corrected x-axis positions, and thereby optical depth.

S **..

S

In one embodiment, step (ii) includes the procedure of Appendix B. 1, thereby producing first and second outputs (Output 1, Output2) dependent upon the detection signals, wherein Outputl is the sum of all correlations between vectors in a first array, the vectors in the first array comprising, for each sweep channel, a stored signal above a threshold that are derived by Direct Fourier Transform from the detection signals, and Output2 is the sum of integrated signals above the threshold for each sweep channel.

In another embodiment, step (ii) includes the procedure of Appendix 8.2, thereby producing a third output (Output3) dependent upon the detection signals, wherein Output3 is, for each sweep channel, the best (lowest) correlation value between the ideal response for a number of barrel lengths and for a number of weapon calibres stored in memory and the direct untransformed detection signals IERI2.

In another embodiment,the detection apparatus includes a first detection element directed in a first direction, towards the entity, for generating non-polarised detection signals, and a second detection element, directed at 90 degrees to the first element, for generating cross-polarised detection signals, and wherein step (ii) includes the procedure of Appendix B.3, thereby producing fourth and fifth outputs (Output4, Output5) dependent upon the detection signals, wherein Output4 is, for each sweep channel, the sum of correlations between the transformed signals, the transformed signals comprising a Complex Fourier Transform of the non-polansed detection signals and of the cross-polarised detection signals, and Output5 is, for each sweep channel, the sum of the integrated non-polansed and cross-polansed signals after Complex Fourier Transform. Preferably, the integrated non-polarised and cross-polarised signals comprise integrations of transformed signals within one or more distance windows, the contents of each distance window being stored in a database in association with a respective response for a particular weapon.

Preferably, the method further comprises provi�ing a neural network, the neural network s. having as inputs thereto any combination of (a) the first and second outputs (Outputl, Output2), (b) the third output (Output3), (c) the fourth and fifth outputs (Output4, Output5), *** and (d) the sixth and seventh outputs (Output6, Output7), wherein the output of the neural * .. network is an indication of a confidence level of a metallic or dielectric object of a predetermined type being detected, for example, I =gun detected, 0 no metallic or dielectric object detected. * *.* I * *

According to another aspect of the present invention there is provided a recorded or recordable medium having recorded or stored thereon digital data defining or transformable into instructions for execution by processing circuitry, the instructions corresponding to at least the steps of any of claims 14 to 26 of the appended claims.

According to another aspect of the present invention there is provided a server comprising processing circuitry, memory and a communications device, the server being programmed for communicating on demand or otherwise digital data defining or transformable into instructions for execution by processing circuitry, the instructions corresponding to at least the steps of any of claims 14 to 26 of the appended claims.

Techniques according to embodiments of the invention entail actively illuminating an object with wide range stepped microwave or millimetre wave radiation and inducing a local electromagnetic field on the surface of the object and within the barrel, comprised of a superposition of modes.

The coupling to these modes from the illuminating and scattered fields is, in general, frequency dependent and this forms the basis for the detection and identification of conducting items. The object needs to be fully illuminated if a full spectrum of modes and therefore a full frequency response are to be excited and collected.

The scattered EM power is typically measured at TMstand off' distance of several metres as the illuminating field is frequency swept over as wide a range as possible, and patterns in frequency response characteristic to the target object being sought are looked for.

This system relies on contributions from one or more of the following effects.

(1) Swept reflectrometry, in which the distances between corners, edges and cavities on the weapon independently from the distances to the source and the detector are ascertained in real-time.

(2) Barrel tone detection. The calibre and barrel length of the weapon can be ascertained when the barrel is oriented towards the general direction of a detector and source (� 450).

This is determined by an aspect invariant method outlined in the appropriate section below.

(3) Cross-polarisation effects. The use of two or more detectors, the first oriented in the same direction as the illuminating radiation and the second at 900 (the cross polarized detector) can yield important effects. The time dependent responses in the cross polarized S...

detector are enhanced when a handgun is present Furthermore, if the frequency is swept over a large range, the reflections from various corners of the object can be resolved in the time domain. These reflections are different or anti-correlated in the cross polarized detector.

This differentiates the technique from previously known polarisation-based detectors outlined above. S. *

(4) Aspect in dependent effects (Late Time Responses). A late time response or I.....

* recurrence from the object is identified, to effectively give information related to the size and dimensions of the object. This enables dimensions to be determined in truly aspect independent manner. This phenomenon is is the further set of scattered signals obtained after the main radiation has returned to the receiver, assuming the application of a pulse of radiation. It is caused by locally induced currents in the object re-radiating. The timing of the response can be directly linked to the size and natural resonances of the object, and this is illustrated herein by simple objects with one or more dimensions. It can, therefore, prove a useful tool in the identification of weapons, optionally when combined with the other techniques identified herein.

The techniques identified in this application are suitable for a deployable gun and concealed weapons detection system and does not rely on imaging techniques for determining the presence of a gun. One or more of the techniques can be combined to reduce the false-positive events from the detector. Hereinafter, experimental sets of responses from typical metal or partially conducting objects such as keys, mobile phones and concealed handguns at a range of frequencies are presented.

An advantage of the invention is that effective detection of threat objects at stand-off distances may be accomplished.

A further advantage of the invention is that detection of threat objects using a portable device may be achieved.

A further advantage is that the detection of smaller dimensions (e.g. gun barrel dimensions and/or calibre) is enabled.

Brief descnotion of the figures Embodiments of the invention will now be described in detail, by way of example, with reference to the accompanying drawings, in which: Figure 1 s a block diagram of the object detection system accorthri9 Lu a first embodiment of * *. the invention; Figure 2(a) shows a block diagram of the object detection system according to a second embodiment of the invention, and Figs 2(b) and 2(c) show the different behaviour of smooth * *. and jagged conducting targets in response to linearly polansed microwave radiation; * Figure 3 shows the original data (amplitude vs frequency) for the radiation detected by the * system of Fig. I from the scan of an x-band waveguide of major dimension 80mm; FIgures 4(a) and 4(b) illustrate respectively Burg and Fourier transforms of the data set of Fig. 3; Figure 5 shows a sample metal object (an X band waveguide connector) with dimensions, as used in testing the system of Fig. I to generate the data of Fig. 3; FIgure 6 shows a comparative plot of the Fourier Transformed data taken from a repeated rapid scan between 14 to 40 Ghz at a standoff distance of a person with small handgun was strapped to the front of a person, as the person rotated round slowly within the beam, using the system of Fig 1 or 2(a), and using swept reflectrometry techniques according to embodiments of the invention; Figure 7 illustrates the same plot as in Fig. 6, but with the gun absent; Figure 8(a) shows a plot using the Burg transform for a scan of a person with simulated plastic explosive strapped to their midriff, using swept reflectrometry techniques according to embodiments of the invention, and Fig. 8(b) shows the same plot as in Fig. 8(b), but with the simulated plastic explosive absent; Figure 9 illustrates measured frequency response of detected signal from a metal barrel with axis aligned at the angles shown to the microwave propagation direction, using barrel tone detection techniques according to the embodiments of the invention; Figure 10 shows correlation of the observed signal with the calculated chirped response function for three different values of f0 and the direct Fourier transform, using barrel tone detection techniques according to the embodiments of the invention; Figure 11 shows representative signals, displayed in the time domain, from a number of targets -in each case the signal has been measured simultaneously with separate horns in two polarisations, using cross-polarisation detection techniques according to embodiments of the invention; Figure 12 illustrates the scanning of a simple metal plate using later time response techniques according to embodiments of the invention, with the EM field polarized in the direction of the 5 cm dimension; Figure 13 shows the time domain response of the 5 cm wide fiat sheet of Fig 12, rotated through vanous angles between 0 and 60 degrees, using later time response techniques according to embodiments of the invention; Figure 14 shows a typical late time responses from a concealed handgun, including the cross-polarised signals (crossed line), using later time response techniques according to embodiments of the invention; and Figure 15 shows schematically the use of a neura! network to make decisions on whether a * i., threat object is present, based on inputs determined by techniques according to embodiments of the invention. ***

* ** Detailed descnotion of embodiments * *

S S.. .

* In the description and drawings, like numerals are used to designate like elements. Unless * * * indicated otherwise, any individual design features and components may be used in combination with any other design features and components disclosed herein. * .

Embodiments of the invention may be used for remotely detecting the presence and size of metal or dielectric objects concealed underneath clothing. This is of interest to the police services, where the detection of concealed handguns and knives is a priority. The invention mainly concerns detection of metal objects, but aspects of this invention may be applied to dielectric objects also. A dielectiic in this context is a non-conducting (i.e. insulating) substance such as ceramic that has a low enough permittivity to allow the microwaves to pass through. A ceramic knife or gun, or a block of plastic explosive, are examples of this type of material.

Two embodiments of the detection system hardware are disclosed herein: Fig. 1 (direct detection without phase detection) and Fig. 2 (detection with phase detection). In each case, the hardware may be embodied in a portable and covertly deployable system Figure 1 is a block diagram of the threat object detection system according to a first embodiment of the invention. For the direct detection (without the phase) responses this embodiment of the detection system optionally includes a microwave or mm wave source 102 (40 GHz Agilent Microwave Synthesizer), although this may comprise a separate component of the system. In this embodiment, the detection system consists a controller (PC) 104, two 20dB standard gain horns used as transmitter 106 and receiver 108 for the Ku and 0 bands, a zero-bias direct detector 110 followed by a DC amplifier 112, and a high speed data acquisition card (PCI-61 32 National Instrument interface). The system is controlled with Labview or C# code, but any appropriate control software may be used. Free wave directional antennae may also replace horns to permit a wider scanning range.

The main procedure carried out by controller 104 in implementing a scan is set out in Appendix A at the end of this disclosure. The procedure Perform transformation on received radiation signals to produce time time domain or optical depth domain trace" in Appendix A, for each of the four techniques, is discussed later with reference to Appendix B. The system typically works by using electromagnetic radiation in the microwave or rni!!imetre * * wave band, where the wavelength is comparable or shorter than the size of the object 116 to * . * be detected. The object 116 may be on or in the body of a person, withing containers or items of luggage, or concealed in or on some other entity (not shown). The suspect entity * (e.g. a person; not shown) has radiation directed by transmitter 106 onto it, so that the (threat) * * object 116 is entirely illuminated by a continuous wave of this radiation (i.e. the radiation is not pulsed, but kept continuously on). The radiation intensity is well within safe operating limits, but is in any case determined by the sensitivity of the detector 110. As an example, in the * : range 14-40 GHz, 0 dBm of power is used with a typical beam area 118 of 0.125m2, which equates to a 20 cm diameter beam. However, it will be appreciated by persons skilled in the art that the hardware may be designed so as generate a beam area 118 of greater or lesser size.

The frequency and consequently the wavelength of the radiation, is swept through a reasonable range (often known as swept CW or continuous wave radiation). Limits are set by the devices used, but a 20 GHz or more sweep starting at 14, 50 or 75 GHz is typical. The data is collected in a series of small steps or as a real-time continuous sweep. Typically 256 or more data points are acquired. In the figures presented in this disclosure, all data is taken between 14 to 40 GHz, providing a sweep range of 26 Ghz.

The illumination and detection is undertaken remotely from the object 116 in question, typically at a distance of a metre or more although there is no lower or upper limit on this distance. The upper limit on detection distance is set by the millimetre or microwave focussing optics, although with this technique a small beam at the diffraction limit is not necessary. The practical ranges for this device are a few tens of cm to many tens of metres with an optimal range of approximately 1 m to 10 m depending on the frequency chosen (some microwave frequencies are attenuated by the atmosphere, and atmospheric windows such as that found around 94 GHz are generally chosen to minimise these effects). Typically the source of electromagnetic radiation 102 and the detector 110 will be mounted next to each other and they will be focussed onto some distant object 116 or entity (not shown).

For techniquel (swept reflectrometiy) and technique2 (barrel tone detection), the return radiation is detected and its amplitude stored as a function of frequency, and the system of Fig. I may be used.

For technique3 (cross-polarization) and technique4 (LTR recurrences), the phase of the returned radiation is acquired at eath frequency point also. When the phase of the returned signal is required -in order to replicate a Time Domain response via the use of a Fourier Transform -the synthesiser and detection system in Fig. I is replaced by a four port Vector Network Analyser (for example 40 GHz Rohde Schwarz SVA Vector Network Analyzers) or another device capable of obtaining the phase and amplitude of the returned signal, once again swept from 14 to 40 GHz in small steps.

FIgure 2(a) shows a block diagram of the object detection system according to a second * ** embodiment of the invention. This indicates the components of the VNA necessary for this embodiment. This is the same as the first embodiment, except as described below. S..

S

The frequencies of the first and second microwave sources 102 and 103 are swept under .. : control of the signal from the Ramp Generator 202 to remain approximately 100 MHz apart.

* The Microwave mixers 204, 206 generate signals corresponding to the difference frequency between the two inputs (-100 MHz). After amplification by two RF amplifiers 208, 210, a RF Mixer 212 produces two outputs corresponding to the "in phase" (I) and "in quadrature" (Q) components of the detected signal. The signals are ampfied by amplifiers 112, 112', and the -11 -data acquisition is controlled by a controller 104 (PC). The entire system apart from the horns 106, 108 is conventionally known as a Vector Network Analyser.

The return signal is collected by a horn and applied to port 2 of the VNA, which measures parameter S21. If cross-polarization measurements are used, a second receiver horn oriented (not shown) at 900 is added to the VNA on (for example) port 3. The transmitted signal is generated from port 1 and is typically 1mW. The real and imaginary parts are recorded and can be corrected for the electrical behaviour of the horns. The signals are zero padded out to 4096 points are processed by a Fast Fourier Transform routine to yield the effective time response. This well known process allows the replication of the application of a pulse of radiation to the target (entity) and the subsequent acquisition of the time resolved response.

Figs 2(b) and 2(c) show the different behaviour of smooth and jagged conducting targets in response to linearly polarised microwave radiation. A single reflection from a metal surface produces a signal polarised in the same manner as the incident signal -polarisation is conserved (Fig. 2(b). However multiple reflections onlaround the target will rotate the direction of polarisation (Fig. 2(c)). This can be preferentially detected by using an additional waveguide horn (not shown) disposed rotated through 90 degrees relative to the first horn 106, thus blocking the polarisation conserving signal.

It will be appreciated by persons skilled in the art that, for each of the embodiments of Figs I and 2(a), or otherwise in accordance with the invention, the hardware thereof may form or be part of a portable device, i.e. small enough to be carried by one person, or transported in an automobile, so as to be operable therein.

Theoretical basis.

To detect the range of siguiais described herein, several measurement techniques are * S. available. For a transmitted signal E0e°' the return signal ER from a target (entity) distance *s..

I'. 1. away may be written as follows: * * * S ER = rEoej°'e2j°' (I) S. I. S : *.: where w=2Trf f is the frequency, r the scattering coefficient and c the velocity of light A detector 110, which responds only to the microwave power, will only measurel ERI 2 which for a single scattering centre, does not explicitly depend on frequency. However for two scattering segments at different ranges L1 and L2, the power is proportional to: IERI2 = f r1e2ftI1 Ic + r2e20 (2) This contains terms in cos(2w(L1 -L2)/c), i.e. oscillatory terms dependent on the difference in range L1-L2. By performing a Fourier transform on the detected power measured as a function of transmitted frequency, peaks corresponding to the difference in range of various parts of the target are observed and these give an indication of the size of the object 116. It is apparent however that for a complicated object 116, many pairs of distances would be involved and the analysis of the signal would be complex.

For a different group of detectors (i.e. Vector Network Analysers; and the embodiment of Fig. 2(a)) it is possible to measure the complex return signal directly. These Vector Network Analysers effectively mix the return signal with a fraction of the transmitted wave to measure r1e2b0)l1c +r2e in terms of its real and imaginary parts. In this case Fourier Transforming leads to a series of signal peaks at the range of each element of the target and arranged in the order of their distance. Thus a much clearer indication of the dimensions of the object 116 is obtained, though only in one dimension. Any Late Time Responses (i.e. those that cannot be attributed to direct scattering) can be measured in this way, although their strength is usually many times less than directly reflected signals.

Further information about the target (entity) may be obtained using a second detector to collect return signals emitted at a different angle from the target. This effectively probes the target along a second directior and can in prindpe enable more dimensions of the object 116 to be ascertained. ** * . S.

A Fourier Transform or some other more advanced power spectrum analysis technique such S..

as a Burg Transform is then applied. The Burg and related methods of power spectrum * analysis have proven to be better than the FT for this application as the individual peaks that * relate directly to the dimensions of the object are more clearly identifiable, as it is possible to choose the number of peaks to be displayed in the output (and hence reject weaker peaks).

. They also allow two closely spaced peaks to be resolved.

S

* FIgure 3 shows the original experimental data (amplitude vs frequency) for the radiation detected by the system of Fig. I from the scan of an x-band waveguide of major dimension 80mm. Figures 4(a) and 4(b) illustrate respectively Burg and Fourier transforms of the data set of Fig. 3. Figure 5 shows a sample metal object (an X band waveguide connector) with dimensions, as used in testing the system of Fig. 1 to generate the data of Fig. 3. The Fourier Transform (Fig. 4(b)) and Burg Transform (Fig. 4(a)) are presented for comparison. The peak at 80 mm corresponds with the length of the waveguide. The Burg algorithm is identified in Fig. 4(a) is much less cluttered than a conventional FT. The Burg algorithm is used here to turn the frequency sweep into a power spectrum, where the peaks in the power spectrum directly give the various dimensions or lengths of the metal or dielectric object, after appropriate scaling.

The position L of peaks within the FFT or Burg spectrum directly relate to the size of the object using the following formula: L = 2Af Where c is the speed of light, tf the penodicity in the frequency domain. This axis will be known as optical depth for the purposes of this disclosure.

The minimum spatial resolution d is related to the sweep range or bandwidth BW: d= C 2BW As an example, if the source frequency were to be swept between 14 and 40 3Hz, this constitutes a sweep range of 26 GHz. This translates to a resolution of 5.7 mm. A larger sweep range would lead to an improved resolution. For example, this will lead to a maximum optical depth or distance of 740 mm for 256 data points. It is straightforward to increase the number of data points to 512, 1024 or any multiple thereof. For Vector Network Analysers * operating in Time Domain mode, in which the complex data is converted to the time domain, this calculation is built into the software.

The four techniques mentioned above, each of which may be used in embodiments of the * . * invention, will now be discussed in more detail.

S S..

Tech nkiuel: SweDt reflectometrv. iSSS * S *5 S

: As briefly described above, swept reflectometry is the principle by which the distances between corners, edges and cavities on the threat object (weapon) 116 independently of the distances to the source (TX horn 106; Fig 1) and the detector (RX horn 108; Fig 1) are ascertained in real-time. Here, I ERI 2as described above is the measured quantity. The distance to the target (entity) is not always available.

If the object 116 (e.g. a weapon strapped to the body as the latter rotates) is then moved around in the beam and its angle and distance with respect to the source and detector is changed, those dimensions between edges and corners that actually belong to the object can be differentiated from those that do not -i.e. the background clutter can be removed.

The main procedure carried out by controller 104 (Fig.1) in implementing a scan is set out in Appendix A at the end of this disclosure. The procedure Perform transformation On received radiation signals to produce time domain or optical depth domain trace" for techniquel is set out in Appendix B.1 at the end of this disclosure.

The software according to techniquel differentiates those peaks that relate to the dimensions of the object 116 from those that do not by acquisition of the signal over a period of time, and by storing these acquired signals independently, with the object moving within the beam. The signals that indirectly relate to the dimensions of the object remain and occur within certain bands denoted by the distances between the various corners of the object, but other signals that change (e.g. the air gaps between clothing and the skin are more chaotic and are integrated out over a period of time).

If the strength of the signal is normalized to distance the returns from a subject conceaHng a handgun will be larger in amplitude than those without.

Figure 6 shows a comparative plot of the Fourier Transformed data taken from a repeated rapid scan between 14 to 40 (5hz at a standoff distance of a person with small handgun strapped to the front of a person, as the person rotated round slowly within the beam, using the system of Fig I or 2(a), and using swept rellectrometry techniques according to embodiments of the invention. FIgure 7 illustrates the same plot as in Fig. 6, but with the gun absent. The scans are presented in three dimensions, with the scan number on the X axis, optical depth on the V axis and the amplitude of the power spectrum (arbitrary units) on the Z axis. In Fig. 6, many resonances can be seen below 100mm with the gun present, denoting * *. the various distances between corners and edges of the weapon. The second plot (Fig. 7) shows the response from the body alone under nearly identical conditions, i.e. the same standoff distance and with the body rotating in the same manner. *

: Very large dimensions such as metal doors, window frames and the multitude of other metal objects will not be observed as they are not entirely encompassed by the beam, as the microwave beam can be focused onto relevant parts of the person (entity) or object 116 in question. -15-

The reflected return radiation is seen to contain patterns or frequencies that can be indirectly related to the dimensions of the metal object according to the technique identified in the section "Theoretical basis" above, including the presence and length of gun barrels, which is in itself can be used to discriminate between, say, handguns and keys, knives and keys, etc. In effect, the technique measures the distances between the various edges of the object at the orientation of the source and detector, and cavity lengths if present The dimensions of guns and knives are different from most other objects carried about the person, so the appropriate dimensions are stored on a database.

The technique (techniquel) is also capable of measuring, particularly, the depth of a dielectric (i.e. of a material that does not conduct electricity) although the physics behind this is significantly different. A dielectric object might typically be a lump of plastic explosives concealed on a suicide bomber.

Figure 8(a) shows a plot using the Burg transform for a scan of a person with simulated plastic explosive strapped to their midriff, using swept reflectrometry techniques according to embodiments of the invention, and Ag. 8(b) shows the same plot as in Fig. 8(b), but with the simulated plastic explosive absent. The Burg transform (Fig. 8(a)) is of a 14-40GHz scan of a person carrying a 80mm thick block of plastic explosive simulant with a dielectric constant of 1.5, gMng an apparent optical depth of 120 mm.

Techniaue2: Barrel tone detection by direct detection of aspect-Independent chlred signals.

Threat objects that contain cavities and can be excited by an incoming microwave signal, can exhibit strong frequency dependence in the scattered signal. These signals differ from those derived from the outside of the object by: *.

1. Showing a threshold frequency for stimulation (cut-off), * 2. Being less dependent on alignment of the cavity with respect to the microwave direction.

S *5*

* For example, consider a 10cm long cylindrical metal barrel closed at one end, which has 19mm outside diameter and 9mm inside diameter. The HI I mode has the lowest threshold frequency fo for propagation for the inside bore at 19.5 GHz. For a cavity length L, the S....

* response for reater than f0 is proportional to the chirped sine wave signal: IERIZcos(2,,JV2 1 2)(2L/c)+q,) (3) where fls the microwave frequency and c is the velocity of light, with a minimal return at lower frequencies below this threshold.

The main procedure carried out by controller 104 (Fig.1) in implementing a scan is set out in Appendix A at the end of this disclosure. The procedure "Perform transformation on received radiation signals to produce time domain or optical depth domain trace" for technique2 is set out in Appendix B.2 at the end of this disclosure.

Figure 9 illustrates measured frequency response of detected signal from a metal barrel with axis aligned at the angles shown to the microwave propagation direction, using barrel tone detection techniques according to the embodiments of the invention. As shown in Figure 9, the return signal clearly shows the onset of the oscillatory response above the threshold f0, i.e., the onset of chirped oscillations occurs when the microwave frequency is beyond cut-off for propagation through the bore of the barrel. The bottom trace shows the optimum calculated chirped response.

It should be noted that the signal is clearly seen at a range of impact angles 0 ranging from 0 to over 450 and the oscillation frequency is only a function of the cavity length L This may be contrasted with the case of edge or corner detection where the oscillation frequency is proportional to LCosO. The analysis provides a means of determining both the length and diameter of the cavity bore. Figure 9. Measured frequency response of detected signal from a metal barrel with axis aligned at the angles shown to the microwave propagation direction.

FIgure 10 shows Correlation of the observed signal with the calculated chirped response function cos(2Tr((2-f?)1'(2Uc)+ço) for three different values of fo and the direct Fourier trar,sform, Ic. =0, using barrel tone detection techniques according to the embodiments of the invention. In Figure 10 the measured signal is correlated with the chirped wave response for particular values of threshold frequency fo. The correlated signal clearly peaks at the ::* actual length L, i.e. 10cm. It can be seen that this signal is much sharper than the direct FT, i.e. when f0 = 0 and the signal is strongly dispersed, i.e. the correlation function is sharper and more symmetric when f0 is equal to the true threshold frequency of 19500 MHz. It is also significantly narrower and more symmetric than when the assumed value for f0 is higher or lower than the correct value. This provides a method of determining both the diameter and length of a cylindrical cavity and thus to identify a threat object such as a gun barrel.

S 5** * S

Technlaue3: IdentificatIon of the taruet by cross-polarised detectors. -17-

By measuring the return signal as a function of frequency scanned over a wide range, the dimensions may be recovered through Fourier Transform techniques. This duplicates the effect of responses of targets to a very short excitation pulse without the need of high speed switches and ultra-fast detectors and digitization processes. The range resolution obtainable is of the order of 0.5-1.0 cm, as described by the principles above, at the sweep ranges available here (14-40 GHz, but this property is not restricted to this frequency range), and thus appropriate for characterising objects such as hand guns.

Another useful aid to threat object identification is to measure the polarisation of the return signal. Waveguide horns (see Fig. 2(a)) act as excellent polarisers and if the transmit 106 and receive 108 horns are similarly orientated, then polarisation conserving components are detected. However, a second horn rotated about its axis by 90 degrees will be blind to *1normal polarisation conserving signals and only detect ucros or polansation changing signals. Now, conducting materials with a smooth surface, including the human body, are mainly polarisation conserving. However complicated targets which involve multiple reflections at different angles and particularly metal objects with sharp edges, give rise to significant "cross polarised signals and this can lead to good discrimination, for example of a hand gun next to the body.

The main procedure carried out by controller 104 (Fig.2(a)) in implementing a scan is set out in Appendix A at the end of this disclosure. The procedure "Perform transformation on received radiation signals to produce time domain or optical depth domain trace" for technique3 is set out in Appendix B.3 at the end of this disclosure.

The responses of a range of objects using the system (Fig. 2(a)), with a second according to the second embodiment, with a cross-polarised detector (not shown) placed almost hiirsledidteiy above the first detector. The test targets ranged from a human alone, front and * side, the same configuration with a concealed small handgun, a bunch of keys, a mobile phone and a digital camera.

Figure 11 shows representative signals, displayed in the time domain, from a number of * targets -in each case the signal has been measured simultaneously with separate horns in two polansations, using cross-polarisation detection techniques according to embodiments of the invention. Figure 11 shows the relevant section of the responses, in which distance and * object information before and after the target has been removed. In each case the signal has *.*** * been measured simultaneously with separate horns in two polansations, parallel to the transmitted beam (single line) and at right angles (crossed line). The quoted range is only relative, typically the targets are 1-2m away from the horns. Signals (a) and (b) are from a -18-small hand gun, (c) and (d) from the chest area of a human body in different orientations. (e) and (f) are for the gun held next to the body, (9) for a small camera held and (h) for a set of keys. It can be can be seen that quite distinctive behaviour is found for the gun in the two potarisations when compared to the body alone and with objects with flat surfaces.

In accordance with technique3, the very wide sweep range leads to detailed information about the dimensions of the object 116 and the fine structure are distances between the source/detector 106/108 and corners/edges on the target/object 116. It can be clearly seen in Fig. 11(a) that the signal from the cross-polarised detector (crossed tine) is not only enhanced by the presence of the weapon but the structural detail is to some extent anti-correlated.

Compare this plot to the one taken from the body-only, where the cross polarized detector shows little information. Similarly, when the gun is placed on the body (either in front or at the side), the cross-polarized signal is once again enhanced, with a degree of anti-correlation between the normal (solid line) and cross polarized (crossed line) detectors.

Technlciue4: Tareet Determination by aspect Independent effects (Late Time Responses).

The Late Time Response (LTR) and the closely associated Singularity Expansion method (SEM) arose from the observation that the time-resolved radar signature from conducting objects contains originating after the radar pulse has passed the target; see The singularity expansion method and its application to target identification, Baum, C.E.; Rothwell, E.J; Chen, K.-M.; Nyquist, D.P. Proceedings of the IEEE Volume 79, Issue 10, Oct. 1991 Page(s): 1481 -1492 arid SEM representation of the early and late time fields scattered from wire targets, Richards, M.A. Antennas and Propagation, IEEE Transactions on, Volume 42, Issue 4, Apr. 1994 Page(s):564 -566. * p *

The pulse sets up currents on the surface of the object 116 in the form of resonant modes.

which subsequently re-radiate. An alternative interpretation is to consider the radar pulse stimulating travelling waves on the surface of the target, which move across and around the object 116 until they return back to their initial distribution. This recurrence can re-radiate . back into the return beam, which appears an extra, time-delayed signal. S. *

An important feature of the LTR is that the time taken for the travelling wave to circulate * around the object does not depend on the orientation of the entity/object 116, and hence is aspect-independent, although its strength may depend on the efficiency of coupling into the modes. For objects 116 with symmetry the time-delay is approximately half the perimeter with respect to centre of the object 116. For more complicated objects, the LTR signal is more complex, but the structure can be interpreted in terms of the dimensions of the object 116. These include dimensions cross range as well as the usual along range values.

Figure 12 illustrates the scanning of a simple metal plate 1202 (PCB) using later time response techniques according to embodiments of the invention. The return signal from a rectangular piece of copper coated PCB was measured. Its narrow dimension (50mm) is orientated in the plane of the microwave electric field and the long dimension (200mm) is at right angles to the polarisation direction and the direction of propagation.

The return signal from return signal is collected by a second horn close to the transmitting horn is measured and applied to port 2 of a VNA 1204 (although the embodiment of Fig. 2(a) may be used), which also provides the transmitted microwaves from port 1. The S21 parameter is recorded over a range typical range 14-40GHz. The signals were corrected for the performance of the microwave horns 106, 108.

If the plate 1202 of width L is rotated by angle 9 about its long direction, then scattering from its leading and trailing edge leads to a doublet response with separation Lsin 9.

The main procedure carried out by controller 104 (Fig.2(a)) in implementing a scan is set out in Appendix A at the end of this disclosure. The procedure "Perform transformation on received radiation signals to produce time domain or optical depth domain trace" for technique4 is set out in Appendix B.4 at the end of this disclosure.

Figure 13 shows the time domain response of the 5 cm wide flat sheet of Fig 12, rotated through various angles between 0 and 60 degrees, using later time response techniques * according to embodiments of the invention. The scattering from leading and trailing edges is centred around 10 cm. There is also a clear Late Time Response at 15 cm whose amplitude is largely independent of angle. Also seen in Fig. 13 is a LTR response at a fixed distance of 5cm. * * *

Figure 14 shows a typical late time responses from a concealed handgun, including the cross-polarised signals (crossed line), using later time response techniques according to embodiments of the invention. This shows the decaying oscillatory responses after 30 cm, being typical late time responses from complex objects. The cross-polarised signal is * * denoted by the crossed line. -20-

Figure 15 shows schematically the use of a neural network 1500 to make decisions on whether a threat object is present, based on inputs 1502 determined by techniques according to embodiments of the invention. The outputs 1504 from the one or more of the abovementioned techniques I to 4 may be taken to the neural network classifier (e.g. a back propagation feed forward network) that has been pre-trained on sets of the type of concealed threat objects 116 that will be of concern, in addition to harmless items.

The training data is preferably formed of sets of data taken using the above-described techniques, in random order. The output 1506 from the neural network is a single output that gives a confidence level (1=gun, 0=no object being concealed). However, it will be appreciated by persons skilled in the art that other configurations or learning algorithms may be used. For example, multiple outputs from the neural network 1500 may be employed, with sub-classifications (e.g. gun, mobile phone, keys, etc).

In conclusion, the millimetre wave reflected signals from the guns show a number of features which enable them to be distinguished from innocently carried objects such as keys and mobile phones. These include cavity mode oscillations from the barrel with characteristic frequency of onset that allows the calibre and length to be determined. The interference of signals from different parts of the target leads to a frequency-dependent response which can be used to deduce the size of the object. The response at different polarisations gives an indication of the complexity of the object from multiple reflections. The late time response gives an aspect-independent signal dependent on target dimensions. Taken together these features provide a means of detecting hand guns under practical conditions at stand-off distances. The application of signal processing techniques enable relevant parameters to be extracted for use in automatic detection systems. ** * *

S

S S.. * I. S

S S

-21 -Appendices Aooendb A Procedure for collection and automatic analysis of threat object sensor Receive activation signal (e.g. operator activates scan button or subject triggers scan] For a predetermined number of sweeps do While full frequency range not scanned do Illuminate subject with radiation Step over frequency range Receive reflected radiation signals Perform transformation on received radiation signals to produce time domain or optical depth domain trace Store in a sweep channel End While Increment sweep channel End do Normalise time domain trace according to range Use Complex Fourier Transform (VNA mode) or Direct Fourier Transform (reflectometry mode) to convert to x-dimension to determine position of trace peaks.

From x-positions, use conversion factor L_C to determine corrected x-axis (optical depth). *..

Perform transformation on received radiation signals by all the steps below: * * a Ai�endixB.1 a. * : [pseudocode for transformation for Swept reflectrometry] If (Techniquel (Swept reflectrometry)) then -22 -Use Direct Fourier Transformed signals (ft t of IERI 2) For each sweep channel do Set Lower and Upper bands Li L2 for useful optical depths (e.g. 10mm to 150mm depending on weapon size and orientation).

Set Threshold for useful signal level above previously collected values for body alone.

From Li to L2 do Store Signal above Threshold separately in vectors in arrayl Integrate Signals above threshold End do End do For each sweep channel do Correlate adjacent vectors and produce outputl Sum with previous outputl' Sum integrated signals above threshold End do Outputi is sum of all correlations between vectors in arrayl Output2 is sum of integrated signals above threshold for each sweep channel Outputl will be different for gun when optical depths change as subject moves in beam than for block of explosive stimulant which is of a similar thickness from different aspects.

Outputi and Output2 are taken to Neural Network input (see Fig. 15).

AoDendix B.2 Else If (Technique2 (Barrel tone detection)) then * * * [pseudocode for transformation for Barrel tone detection] Use Direct Untransformed signals IERI 2 *. * For each sweep channel do : For set of weapon calibres do For set of barrel lengths do Calculate onset (fO) for calibre Calculate chirped response for Length L -23 -Correlate ideal response cos(2r(P-f02)"2(2Uc)+p) with data.

Store Correlation value End do End do Find best (lowest) correlation value and store in Output3 End do Keep fO and L and display Output3 goes to Neural network (see Fig. 15).

++ ++ + + + ++ + +++ + + Appendix B.3 Else if (Technique3 (Cross-polarisation detection)) then (pseudocode for transformation for Cros-polarisation detection] Use Complex Fourier Transform signals (ER) which give range information, from normal and cross polarised detectors Select Distancel which is first significant reflection above a threshold (or which is given by an independent range finding sensor) For each sweep channel do Apply a distance window of a given number of millimetres determined by database of responses from weapons.

Select trailing edge of response (distance2) by adjusting distance window to when response falls below threshold Integrate response from non-pol detector within window Integrate response from cross-pol detector within window Sum to previous integrations

S

* Correlate response from non-pol detector with cross-pol Store and awn correlation * S End do * Outputs 4 and 5 are sum of correlations and sum of integrations * Outputs are taken to Neural Network inputs (see Fig. 15).

S

S :"

Appendix 8.4 -24-Else if (Technique4 (Late time response detection and resonant frequency detection)) then [pseudocode for transformation for Late time response] Use Complex Fourier Transform signals (Es) which give range information, from normal and cross polarised detectors For each sweep channel do Select Distance2 which is the output of Technique3 Apply a distance window of a given number of millimetres determined by database of late time responses from weapons.

Find the smoothed responses of the late time responses (see Fig. 14) in normal and cross-pol detectors.

Ascertain the exponential decay rate by a non-linear least squares fitting procedure. Output is stored in a vector End do Normalise outputs from the vector and form Output 6 Use Complex Fourier transformed signals (ER), from normal and cross polarised detectors For each sweep channel do For the entire sweep data If response level above a normalised threshold then Apply series of non-linear filters (e.g. MUSIC filtef) with filter characteristics taken from a data base to look for a particular resonance Store the magnitude of the resonance Apply a peak detection algorithm (E.G. zero crossing) Store peak locations * . End if End do For each sweep channel do Compare peak locations with known natural resonances for object Sum the differences between peak locations and natural resonances from data base. p...

: Output7 is sum of differences -to Neural Network.

Keep the peak locations for display, which can indicate weapon type. *

End do End if // End of transformation techniques phase -25-

A

* See A Subspace Signal Processing Technique for Concealed Weapons Detection, Ibrahim, A.S.; Liu, K.J.R.; Novak, D.; Waterhouse, R.B., Acoustics, Speech and Signal Processing, 2007. ICASSP 2007. IEEE International Conference on, Volume 2, Issue, 15-20 April 2007 Page(s):Il-401 -11-404. I * S. *5*

S I. * S

S * S.-55* S. I. S

S

S a I -26-

Claims

Claims: 1. A system for remote detection of one or more dimensions of a metallic or dielectric object, comprising: a transmission apparatus, including a transmission element, for directing microwave or mm wave radiation in a predetermined direction, a detection apparatus, for receiving radiation from an entity resulting from the transmitted radiation and generating one or more detection signals in the frequency domain, and a controller, the controller being operable to (i) cause the transmitted radiation to be swept over a predetermined range of frequencies, (ii) perform a transform operation on the detection signal(s) to generate one or more transformed signals in the time domain, and (iii) determine, from one or more features of the transformed signal, one or more dimensions of a metallic or dielectric object upon which the transmitted radiation is incident.
2. The system of claim 1, wherein the transmission element is a directional element that may be pointed by a user.
3. The system of claim 1 or 2, wherein the controller is operable to initiate step (i) upon receiving an activation signal, the activation signal corresponding to a user input or to detection of the presence of the entity.
4. The system of claim 1, 2 or 3, wherein: step (i) comprises stepwise sweeping by predetermined steps in frequency: and step (ii) comprises (iia) perform a transform operation after each step to produce a time domain or optical depth domain trace, and (iib) storing each time domain or optical depth *
5 domain trace in a respective sweep channel, said time domain or optical depth domain traces S.. thereby comprising said transformed signals. 4.

* *. 5. The system of any of claims 1 to 4, wherein (ni) compnses normalising according to range one or more transformed signals.
6. The system of claim 5, wherein (iii) further comprises using a Complex Fourier 0* Transform or Direct Fourier Transform to convert transformed signals to the x-dimension, and S....* determining the x-positions of peaks on the transformed signals.
7. The system of claim 6, wherein (iii) further comprises, from the x-positions, using -27 - An--L = 2f where I = distance to entity c = the speed of light fr the penodicity in the frequency domain, to determine corrected x-axis positions, and thereby optical depth.
8. The system of any of the preceding claims, wherein step (ii) includes the procedure of Appendix B. 1, thereby producing first and second outputs (Output 1, Output2) dependent upon the detection signals, wherein Outputl is the sum of all correlations between vectors in a first array, the vectors in the first array comprising, for each sweep channel, a stored signal above a threshold that are derived by Direct Fourier Transform from the detection signals, and Output2 is the sum of integrated signals above the threshold for each sweep channel.
9. The system of any of the preceding claims, wherein step (ii) includes the procedure of Appendix 8.2, thereby producing a third output (Output3) dependent upon the detection signals, wherein Output3 Is, for each sweep channel, the best (lowest) correlation value between the ideal response for a number of barrel lengths and for a number of weapon calibres stored in memory and the direct untransformed detection signals IERI2.
10. The system of any of the preceding claims, wherein the detection apparatus includes a first detection element directed in a first direction, towards the entity, for generating non-polarised detection signals, and a second detection element, directed at 90 degrees to the first element, for generating cross-polansed detection signals, and wherein step (ii) includes the procedure of Appendix 8.3, thereby producing fourth *..., and fifth outputs (Output4, Output5) dependent upon the detection signals, wherein S...Output4 is, for each sweep channel, the sum of correlations between the the * transformed signals, the transformed signals comprising a Complex Fourier Transform of the * non-polarised detection signals and of the cross-polarised detection signals, and *.* Output5, for each sweep channel, the sum of the integrated non-polarised and cross- : ::: polansed signals after Complex Fourier Transform.S ***.

*
11. The system of claim 10, wherein the integrated non-polarised and cross-polarised signals comprise integrations of transformed signals within one or more distance windows, the contents of each distance window being stored in a database in association with a respective response for a particular weapon.

-28 -
12. The system of any of the preceding claims, wherein step (ii) includes the procedure of Appendix B.4, thereby producing sixth and seventh outputs (Output6, Output7) dependent upon the detection signals.
13. The system of any of the preceding claims, further including a neural network, the neural network having as inputs thereto any combination of (a) the first and second outputs (Outputl, Output2), (b) the third output (Output3), (C) the fourth and fifth outputs (Output4, Output5), and (d) the sixth and seventh outputs (Output6, Output7), wherein the output of the neural network is an indication of a confidence level of a metallic or dielectric object of a predetermined type being detected, for example, 1=gun detected, 0= no metallic or dielectric object detected.
14. A method for remote detection of one or more dimensions of a metallic or dielectric object, comprising: providing a transmission apparatus, induding a transmission element, for directing microwave or mm wave radiation in a predetermined direction, providing a detection apparatus, for receiving radiation from an entity resulting from the transmitted radiation and generating one or more detection signals in the frequency domain, and providing a controller, and operating the controller to (iv) cause the transmitted radiation to be swept over a predetermined range of frequencies, (v) perform a transform operation on the detection signal(s) to generate one or more transformed signals in the time domain, and (vi) determine, from one or more features of the transformed signal, one or more dimensions of a metallic or dielectric object upon which the transmitted radiation is incident. * ** * * * *.
15. The method of claim 14, wherein the transmission element is a directional element that may be pointed by a user. * ** * 0S
16. The method of claim 14 or 15, wherein the controller is operable to initiate step (i) upon receiving an activation signal, the activation signal corresponding to a user input or to : detection of the presence of the entity.S 0S*0S * 0
17. The method of claim 14, 15 or 16, wherein: step (i) comprises stepwise sweeping by predetermined steps in frequency; and step (ii) comprises (ha) perform a transform operation after each step to produce a time domain or optical depth domain trace, and (iib) storing each time domain or optical depth -29-domain trace in a respective sweep channel, said time domain or optical depth domain traces thereby comprising said transformed signals.
18. The method of any of daims 14 to 17, wherein (iii) comprises normalising according to range one or more transformed signals.
19. The method of claim 18, wherein (iii) further comprises using a Complex Fourier Transform or Direct Fourier Transform to convert transformed signals to the x-dimension, and determining the x-positions of peaks on the transformed signals.
20. The method of claim 19, wherein (iii) further comprises, from the x-positions, using L = 2Af where L = distance to entity c = the speed of light M= the periodicity in the frequency domain, to determine corrected x-axis positions, and thereby optical depth.
21. The method of any of the preceding claims, wherein step (ii) includes the procedure of Appendix B. 1, thereby producing first and second outputs (Outputl, Output2) dependent upon the detection signals, wherein Outputl is the sum of all correlations between vectors in a first array, the vectors in the first array comprising, for each sweep channel, a stored signal above a threshold that are derived by Direct Fourier Transform from the detection signals, and Output2 is the sum of integrated signals above the threshold for each sweep channel.
22. The method of any of the preceding claims, wherein step (ii) includes the procedure of Appendix B.2, thereby producing a third output (Output3) dependent upon the detection ***** signals, wherein * S..Output3 is, for each sweep channel, the best (lowest) correlation value between the ideal response for a number of barrel lengths and for a number of weapon calibres stored in memory and the direct untransformed detection signals IERI2.

: : :
23. The method of any of the preceding claims, wherein the detection apparatus includes a first detection element directed in a first direction, towards the entity, for generating *55*5 * non-polarised detection signals, and a second detection element, directed at 90 degrees to the first element, for generating cross-polansed detection signals, and -30-wherein step (ii) includes the procedure of Appendix B.3, thereby producing fourth and fifth outputs (Output4, Output5) dependent upon the detection signals, wherein Output4 is, for each sweep channel, the sum of correlations between the the transformed signals, the transformed signals comprising a Complex Fourier Transform of the non-polarised detection signals and of the cross-polarised detection signals, and Output5, for each sweep channel, the sum of the integrated non-polansed and cross-polansed signals after Complex Fourier Transform.
24. The method of claim 10, wherein the integrated non-polarised and cross-polansed signals comprise integrations of transformed signals within one or more distance windows, the contents of each distance window being stored in a database in association with a respective response for a particular weapon.
25. The method of any of the preceding claims, wherein step (ii) includes the procedure of Appendix 8.4, thereby producing sixth and seventh outputs (Output6, Output7) dependent upon the detection signals.
26. The method of any of the preceding claims, further including a neural network, the neural network having as inputs thereto any combination of (a) the first and second outputs (Output 1, Output2), (b) the third output (Output3), (c) the fourth and fifth outputs (Output4, Output5), and (d) the sixth and seventh outputs (Output6, Output7), wherein the output of the neural network is an indication of a confidence level of a metallic or dielectric object of a predetermined type being detected, for example, 1=gun detected, 0= no metallic or dielectric object detected.
27. A recorded or recordable medium having recorded or stored thereon digital data defining or transformable into instructions for execution by processing circuitry, the instructions corresponding to at least the steps of ny of claims 14 to 26. * ** * * * **
28. A server comprising processing circuitry, memory and a communications device, the server being programmed for communicating on demand or otherwise digital data defining or transformable into instructions for execution by processing circuitry, the instructions *:. corresponding to at least the steps of any of claims 14 to 26.: : ::
29. A detection system for remote detection of one or more dimensions of a metallic or * dielectric object, the system being substantially as hereinbefore described with reference to * the accompanying drawings.-31 -
30. A method for remote detection of one or more dimensions of a metallic or dielectric object, the method being substantially as hereinbefore described with reference to the accompanying drawings. * *S * * * ** S... * .5* * S. * . p *S S.. *5*S * S * .SS S-32 -