KEYWORDS: Synthetic apertures, 3D image processing, Point spread functions, Antennas, Systems modeling, Imaging systems, Visibility, Receivers, 3D displays, Spatial resolution
This paper presents part of a feasibility study into the use of the aperture synthesis passive imaging technique to screen vehicles for persons. The aperture synthesis technique is introduced and shown how in the near-field regime of a vehicle screening scenario that a three-dimensional imaging capability is possible. A suggested antenna receiver array is presented and the three-dimensional point spread function which this enables is calculated by simulation. This shows that over the majority of the inside of the vehicle the spatial resolution in all three spatial dimensions is of or less than the radiation wavelength, which at the suggested operational radiation frequency of 20 GHz is 1.5 cm. A radiation transport model that estimates the radiation temperatures of persons and backgrounds when viewing the vehicle either from the side or the top is presented, such a model being useful in the design of vehicle screening systems and as a basis for interpretation codes to assist operators in recognising persons in vehicles.
This paper investigates by simulation the use of the three-dimensional aperture synthesis imaging technique to image three-dimensional extended sources. Software was written to access the three-dimensional information from computer graphics models in the formats of *.dxf and *.3ds and use these to generate synthetic cross-correlations, as if they would have been generated by an aperture synthesis antenna/receiver array measuring the radiometric emission from the three-dimensional object. A three-dimensional (near-field) aperture synthesis imaging algorithm generates [1] a voxel image of the three-dimensional object. Images created from a sphere indicate faithful reproduction about a single phase centre when the radius of the sphere is less than the Fresnel scale. However, for larger spheres, definition in the threedimensional imagery suffers and a phenomenon, referred to in this paper as Fresnel noise, appears in the image. Images of objects larger than the Fresnel scale can be created by having multiple smaller images, each having a size approximately of the Fresnel scale and centred on separate phase centres. Using the software to generate threedimensional imagery of a person, to demonstrate capabilities for portal security screening, indicates the technique works to first order. Improvements are needed in the software to improve the spatial sampling of the radiometric fields from the three-dimensional objects and implement a volumetric image mosaicking technique to remove the Fresnel noise.
The threat of concealed weapons, explosives and contraband in footwear, bags and suitcases has led to the development of new devices, which can be deployed for security screening. To address known deficiencies of metal detectors and x-rays, an UWB 3D microwave imaging scanning apparatus using FMCW stepped frequency working in the K and Q bands and with a planar scanning geometry based on an x y stage, has been developed to screen suspicious luggage and footwear. To obtain microwave images of the concealed weapons, the targets are placed above the platform and the single transceiver horn antenna attached to the x y stage is moved mechanically to perform a raster scan to create a 2D synthetic aperture array. The S11 reflection signal of the transmitted sweep frequency from the target is acquired by a VNA in synchronism with each position step. To enhance and filter from clutter and noise the raw data and to obtain the 2D and 3D microwave images of the concealed weapons or explosives, data processing techniques are applied to the acquired signals. These techniques include background subtraction, Inverse Fast Fourier Transform (IFFT), thresholding, filtering by gating and windowing and deconvolving with the transfer function of the system using a reference target. To focus the 3D reconstructed microwave image of the target in range and across the x y aperture without using focusing elements, 3D Synthetic Aperture Radar (SAR) techniques are applied to the post-processed data. The K and Q bands, between 15 to 40 GHz, show good transmission through clothing and dielectric materials found in luggage and footwear. A description of the system, algorithms and some results with replica guns and a comparison of microwave images obtained by IFFT, 2D and 3D SAR techniques are presented.
The three dimensional (3D) aperture synthesis imaging technique investigated here is a generalisation of the classic twodimensional
radio astronomy technique with refinements for the near-field so it can be applied a personnel security
screening portal. This technique can be viewed as a novel form of diffraction emission tomography and extends previous
3D aperture synthesis imaging research using matrix inversion techniques [1]. Simulations using three-dimensional
Fourier transforms to create three-dimensional images from simulated three-dimensional visibility functions illustrate the
Abbe microscopy resolution should be achievable in three dimensions simultaneously in a single sensor. The field-of-view
is demonstrated to be limited by Fresnel scale effects and a means to over coming this by processing sub-sets of
local visibility functions with different phase centres throughout the imaging volume is presented. The applications of
this technique to a full 3D imaging security screening portal is explored and a route to extending simulation software for
market driven imaging scenarios is discussed.
This paper examines the suitability and potential of reducing the acquisition requirements of a novel radiation mapper
through the application of the non-linear deconvolution technique, CLEAN. The radiation mapper generates a threshold
image of the target scene, at a user defined distance, using a single pixel detector manually scanned across the scene .
This paper provides a discussion of the factors involved and merits of incorporating CLEAN into the system. In this
paper we describe the modifications to the system for the generation of an intensity map and the relationship between
resolution and acquisition time for a target scene. The factors influencing image fidelity for a scene are identified and
discussed with the impact on fill-factor of the intensity image, which in turn determines the ability of the operator to
accurately identify features of the radiation source within a target scene. The CLEAN algorithm and its variants have
been extensively developed by the radio astronomy community to improve the image fidelity of data collected by sparse
interferometric arrays. However, the algorithm has demonstrated surprising adaptability including terrestrial imagery, as
detailed in Taylor et al. SPIE 9078-19 and Bose et al., IEEE 2002. CLEAN can be applied directly to raw data via a
bespoke algorithm. However, this investigation is a proof-of-concept and thus requires a well tested verification method.
We have opted to use the public ally available implementation of CLEAN found in the Common Astronomy Software
Applications (CASA) package. The use of CASA for this purpose dictates the use of simulated input data and radio
astronomy standard parameters. Finally, this paper presents the results of applying CLEAN to our simulated target scene,
with a discussion of the potential merits a bespoke implementation would yield.
This paper investigates the application of the CLEAN non–linear deconvolution method to Late Time Response (LTR)
analysis for detecting multiple objects in Concealed Threat Detection (CTD). When an Ultra-Wide Band (UWB)
frequency radar signal is used to illuminate a conductive target, surface currents are induced upon the object which in
turn give rise to LTR signals. These signals are re-radiated from the target and the results from a number of targets are
presented.
The experiment was performed using double ridged horn antenna in a pseudo-monostatic arrangement. A Vector
network analyser (VNA) has been used to provide the UWB Frequency Modulated Continuous Wave (FMCW) radar
signal. The distance between the transmitting antenna and the target objects has been kept at 1 metre for all the
experiments performed and the power level at the VNA was set to 0dBm. The targets in the experimental setup are
suspended in air in a laboratory environment.
Matlab has been used in post processing to perform linear and non-linear deconvolution of the signal. The Wiener filter,
Fast Fourier Transform (FFT) and Continuous Wavelet Transform (CWT) are used to process the return signals and
extract the LTR features from the noise clutter. A Generalized Pencil-of-Function (GPOF) method was then used to
extract the complex poles of the signal. Artificial Neural Networks (ANN) and Linear Discriminant Analysis (LDA)
have been used to classify the data.
The unique selling proposition of millimetre wave technology for security screening is that it provides a stand-off or portal scenario sensing capability for non-metallic threats. The capabilities to detect some non-metallic threats are investigated in this paper, whilst recommissioning the AVSEC portal screening system at the Manchester Metropolitan University. The AVSEC system is a large aperture (1.6 m) portal screening imager which uses spatially incoherent illumination at 28-33 GHz from mode scrambling cavities to illuminate the subject. The imaging capability is critically analysed in terms of this illumination. A novel technique for the measurement of reflectance, refractive index and extinction coefficient is investigated and this then use to characterise the signatures of nitromethane, hexane, methanol, bees wax and baking flour. Millimetre wave images are shown how these liquids in polycarbonate bottles and the other materials appear against the human body.
At airports, security screening can cause long delays. In order to speed up screening a solution to avoid passengers removing their shoes to have them X-ray scanned is required. To detect threats or contraband items hidden within the shoe, a method of screening using frequency swept signals between 15 to 40 GHz has been developed, where the scan is carried out whilst the shoes are being worn. Most footwear is transparent to microwaves to some extent in this band. The scans, data processing and interpretation of the 2D image of the cross section of the shoe are completed in a few seconds. Using safe low power UWB radar, scattered signals from the shoe can be observed which are caused by changes in material properties such as cavities, dielectric or metal objects concealed within the shoe. By moving the transmission horn along the length of the shoe a 2D image corresponding to a cross section through the footwear is built up, which can be interpreted by the user, or automatically, to reveal the presence of concealed threat within the shoe. A prototype system with a resolution of 6 mm or less has been developed and results obtained for a wide range of commonly worn footwear, some modified by the inclusion of concealed material. Clear differences between the measured images of modified and unmodified shoes are seen. Procedures for enhancing the image through electronic image synthesis techniques and image processing methods are discussed and preliminary performance data presented.
KEYWORDS: Imaging systems, Thin film coatings, Signal attenuation, Signal detection, Firearms, Scattering, K band, Antennas, Ka band, Spatial resolution
The feasibility of screening hand luggage for concealed threat items such as Person-Borne Improvised Explosive Devices (PBIED's) both metallic and non-metallic, together with handguns and at millimetre wavelengths is investigated. Previous studies by the authors and others indicate that hand baggage material and fabric is much more transmissive and has less scattering at lower millimetre wave frequencies and the ability to use K-band active imaging with high spatial resolution presents an opportunity to image and hence recognise concealed threats. For this feasibility study, a 1.6 m aperture, 35 GHz security screening imaging system with a spatial resolution of 2.5 cm and a depth of field of around 5 cm is employed, using spatially incoherent illuminating panels to enhance image contrast. In this study, realistic scenarios using backpacks containing a realistic range of threat and non-threat items are scanned, both carried and standalone. This range of items contains large vessels suitable for containing simulated home-made PBIED’s and handguns. The comprehensive list of non-threat items includes laptops, bottles, clothing and power supplies. For this study, the range at which imaging data at standoff distances can be acquired is confined to that of the particular system in use, although parameters such as illumination and integration time are optimised. However, techniques for extrapolating towards effective standoff distances using aperture synthesis imagers are discussed. The transmission loss through fabrics and clothing that may form, or be contained in baggage, are reported over range of frequencies ranging from 26 to 110 GHz.
Target decomposition is of interest to the security and defence community as it enables data sets to be reduced to their principal identifying components as a pre-processor to running pattern recognition algorithms. An investigation into the application of target decomposition theory for concealed threat detection is presented. A validation study has been conducted in which the scattering of an EM wave by a number of primitive radar calibration targets has been simulated using an EM FEA solver. A knife has then been illuminated with quad polar radar again in a simulation conducted using an EM FEA solver. The validation of the target decomposition algorithm is achieved by analysing the calibration targets with well-known scattering mechanisms. The decomposition of more complex targets such as those encountered in Concealed Threat Detection (CTD) scenarios are then analysed. The decomposition is performed as prescribed by Cloude et al and the scattering of the illuminating wave due to the target is mapped onto a 3D space detailing polarimetric entropy (H), anisotropy (A) and alpha-angle (α). The decomposition of these complex scattering mechanisms is then used to classify the data. Both theoretical and simulated data sets are used to validate this technique.
This paper investigates by simulation some of the capabilities of near-field and three-dimensional imaging which are made possible by accessing phase and amplitude of electric fields from radiometric emission using aperture synthesis systems. The aperture synthesis technique is the main stay of high resolution radio astronomy and is investigated here for the near-field application of personnel security screening in the millimetre wave band. The limitations of the standard radio astronomy visibility-function technique and a matrix method for image generation are investigated for this purpose. It is concluded that several hundred receivers are required for high pixel count (> few thousand) and good quality images and that a new and more efficient algorithms is required to process such numbers of channels from non-planar imaging arrays in the near-field. Investigating the resolution limits of three-dimensional imaging in the near-field region with this technique indicates sub-wavelength resolution may be possible
This paper investigates the use of Late Time Response (LTR) analysis for detecting multiple objects in concealed object detection. When a conductive object is illuminated by an ultra-wide band (UWB) frequency radar signal, the surface currents induced upon the object give rise to LTR signals. The LTR results from a number of different targets are presented. The distance between the targets within the same radar beam has been adjusted in increments of 5cm to determine the point at which the individual objects can be distinguished from each other. The experiment was performed using double ridged horn antennas in a pseudo-monostatic arrangement. Vector network analysers (VNA) are used to provide the UWB stepped frequency continuous wave radar signal. The distance between the transmitting antenna and the target object is kept at 50cm for all the experiments performed and the power level at the VNA was set to 2dBm. The targets in the experimental setup are suspended in isolation in a non-anechoic environment. To allow for the de-convolution of the signal and the removal of background clutter Matlab was used in post processing. The Fast Fourier Transform (FFT) and Continuous Wavelet Transform (CWT) are used to process the return signals and extract the LTR features from the noise clutter. A Generalized Pencil-of-Function (GPOF) method was then used to extract the complex poles of the signal. In the case of a single needle these poles can be found around 1.9GHz.
A millimeter wave (75 - 110 GHz) polarimetric radar system (MiRTLE) has been developed for the detection of threat
objects, such as guns, knives, or explosive devices, which have been concealed under clothing upon the human body.
The system uses a Gaussian lens antenna to enable operation at stand-off ranges up to 25 meters. By utilizing ultra-wideband
Swept Frequency Continuous Wave Radar very high range resolution (~ 10mm) is realized. The system is
capable of detecting objects positioned in front of the body and of measuring the range of a target. By interpretation of
the scattered waveform, the presence of a wide spectrum of threat items concealed on the human body may be detected.
Threat detection is autonomously rendered by application of a neural network to the scattered time domain, polarimetric
radar returns and the system may be taught to alarm or reject certain classes of objects; this allows for highly specific or
broad spectrum threat detection. The radar system is portable and operator steerable allowing standoff monitoring of
moving human targets in real time. Rapid (1ms) sweep times and fast signal acquisition and processing allow decisions
to be made at video frame rates (30 fps) and integrated directly to a video feed providing the operator with a field of
view and facilitating aiming. Performance parameters for detection of guns and simulated explosive devices are
presented for ranges up to 25 meters.
A method of effectively detecting remote concealed threats, particularly knives and guns, has been developed. This
method uses multi-polarimetric ultra wide band active microwave radar to remotely scan a person under investigation. It
has been shown that the radar signatures from such scans can be used to detect whether a person is carrying a concealed
threat. A Principal Component Analysis (PCA) data reduction technique followed by a neural network (NN) is used to
classify the information extracted from the radar signals. The technique combines the co, 45°, cross, and 135° polarized
transceived radar signals into a single data set for classification. Illuminating the target with a range of polarizations,
together with choosing a radar beam size commensurate with the targets in question, produces good discrimination
between threat and non-threat items. Once collected, the data sets obtained are reduced via PCA, which significantly
improves the correct classification rate at the NN stage and makes the technique more tolerant of variations in the threat
objects orientation and better able to detect a wider range of threat types. Experimental results are presented which show
that a detection rate of up to 80% for knives and guns can be achieved, with a false alarm rate as low as 4%.
A millimetre wave (75 - 110 GHz) polarimetric RADAR system is demonstrated for the detection of threat objects
concealed under clothing upon the human body at stand-off ranges of up to 25 metres. The system implements Swept
Frequency Continuous Wave RADAR with low cost components to deliver a compact, UWB, high resolution (~ 1 cm)
RADAR system capable of detecting, resolving and discriminating a wide spectrum of threat items concealed on the
human body. Threat detection is autonomously rendered by application of a neural network to the scattered time domain
polarimetric radar return, the system may be taught to alarm or reject certain classes of objects; allowing for highly
specific through to broad spectrum threat detection. The authors present data for some simple envisaged threat scenarios
at stand off ranges out to 25 metres.
A method of detecting concealed handguns and knives, both on and off body, has been developed. The method utilizes
aspect-independent natural, complex resonances (poles) excited by illuminating the target with frequency swept, ultrawide
band microwaves in the range 0.5 - 18 GHz. These natural resonances manifest as a Late Time Response (LTR)
that extends significantly (~ 5 ns) beyond the direct reflections from the human body (the Early Time Response) and are
of the form of a superposition of exponentially decaying sinusoidal waveforms. Two handguns are examined, both on
the human body and in isolation, by the established methodology of applying the Generalised-Pencil-Of-Function to the
late time response data of the target. These poles allow the weapon to be effectively classified. Out of plane polarized
(cross-polarized) scattered response is used here as this gives improved discrimination between the early and late time
responses. Determination of the presence or absence of particular weapons concealed under clothing, on the human
body, is demonstrated. A novel bow-tie slot antenna is described which has good pulse and frequency response over the
range 0.3-1 GHz and which is suitable for excitation of the fundamental natural resonances.
Active millimetre wave systems, operating at frequencies up to 110 GHz have been used to detect the presence of both
concealed dielectric and metallic objects at standoff distances. Co- and cross-polarized superheterodyne or direct
detectors are used to differentiate between metallic and purely dielectric objects. The technique determines the thickness
of a dielectric target and detects the presence of concealed handguns or fragmentation by utilising the pattern of the
responses from both the co- and cross-polarized detectors. The returned signals are processed and analysed by an
artificial neural network, which classifies the responses according to their correspondence to previous training data.
The effective detection of concealed handguns and knives in open spaces is a major challenge for police and security
services round the world. Here an automated technique for the detection of concealed handguns that relies on active
swept illumination of the target to induce both scattered fields and aspect independent responses from the concealed
object is presented. The broad frequency sweep permits information about the object's size to be deduced from
transformations into the time/distance domain. In our experiments we collect multiple sweeps across the frequency range
at very high speed, which produces a time evolved response from the target, from both normal and cross polarized
detectors. From this we extract characteristic signatures from the responses that allow those from innocent objects (e.g.
mobile phones, keys etc) to be distinguished from handguns. Information about the optical depth separation of the
scattering corners and the degree and shape of cross polarization allows a neural network to successfully concealed
handguns. Finally this system utilizes a range of signal processing techniques ranging from correlation between cross
and normally polarized scattering through to a neural network classifier to deduce whether a concealed weapon is
present.
Millimetre waves in the range 20 to 110 GHz have been used to detect the presence and thickness of dielectric materials, such as explosives, by measuring the frequency response of the return signal. Interference between the reflected signals from the front and back surfaces of the dielectric provides a characteristic frequency variation in the return signal, which may be processed to yield its optical depth [Bowring et al, Meas. Sci. Technol. 19, 024004 (2008)]. The depth resolution depends on the sweep bandwidth, which is typically 10 to 30 GHz. By using super-heterodyne detection the range of the object can also be determined, which enables a signal from a target, such as a suicide bomber to be extracted from background clutter. Using millimetre wave optics only a small area of the target is illuminated at a time, thus reducing interference from different parts of a human target. Results are presented for simulated explosive materials with water or human backing at stand-off distances. A method of data analysis that involves pattern recognition enables effective differentiation of target types.
Guns and knives have become a significant threat to public safety. Recently, a variety of techniques based on Electromagnetics (EM) have been used for their detection. For example, walk-through metal detection has been used in airports; X-ray and THz detection systems have been used for luggage screening. Different EM frequencies for metallic object detection have demonstrated different merits. This paper reports on a 1-14 GHz swept-frequency radar system for metallic object detection using reflection configuration. The swept frequency response and resonant frequency behaviour of a number of metallic objects, in terms of position, object shape, rotation and multiple objects have been tested and analysed. The system working from 1 to 14 GHz has been set up to implement sensing of metal items at a standoff distance of more than 1 meter. Through a series of experimental investigations, it can be found that the optical depths derived from the Fourier Transform of the power spectrum profile is in close relation with the relative location of the metallic object. The cross correlation between coherence-polarisation and cross-polarisation RF returns can be used to distinguish different objects. Therefore the optical depth and the cross correlation can be used as useful features for metallic object detection and characterisation in this portion of the microwave frequency spectrum.
An active technique for the standoff detection and identification of concealed conducting items such as handguns and
knives is presented. This technique entails illuminating an object with wide range stepped millimetre wave radiation and
inducing a local electromagnetic field 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 measured at "stand off" distance of
several metres as the illuminating field is frequency swept and patterns in frequency response characteristic to the target
item being sought are looked for. This system relies on contributions from the aspect independent late time responses
employed by Baum1 together with aspect independent information derived specifically from gun barrels and polarisation
from scattering effects. This technique is suitable for a deployable gun and concealed weapons detection system and
does not rely on imaging techniques for determining the presence of a gun. Experimental sets of responses from typical
metal or partially conducting objects such as keys, mobile phones and concealed handguns are presented at a range of
frequencies.
A novel edge detector has been developed that utilises statistical masks and neural networks for the optimal detection of edges over a wide range of image types. The failure of many common edge detection techniques has been observed when analysing concealed weapons X-ray images, biomedical images or images with significant levels of noise, clutter or texture. This novel technique is based on a statistical edge detection filter that uses a range of two-sample statistical tests to evaluate any local image texture differences and by applying a pixel region mask (or kernel) to the image analyse the statistical properties of that region. The range and type of tests has been greatly expanded from the previous work of Bowring et al.1 This process is further enhanced by applying combined multiple scale pixel masks and multiple statistical tests, to Artificial Neural Networks (ANN) trained to classify different edge types. Through the use of Artificial Neural Networks (ANN) we can combine the output results of several statistical mask scales into one detector. Furthermore we can allow the combination of several two sample statistical tests of varying properties (for example; mean based, variance based and distribution based). This combination of both scales and tests allows the optimal response from a variety of statistical masks. From this we can produce the optimum edge detection output for a wide variety of images, and the results of this are presented.
Extensive work has been published on millimetre wave active and passive detection and imaging of metallic objects
concealed under clothing. We propose and demonstrate a technique for revealing the depth as well as the outline of
partially transparent objects, which is especially suited to imaging layer materials such as explosives and drugs.
The technique uses a focussed and scanned FMCW source, swept through many GHz to reveal this structure. The
principle involved is that a parallel sided dielectric slab produces reflections at both its upper and lower surfaces, acting
as a Fabry-Perot interferometer. This produces a pattern of alternating reflected peaks and troughs in frequency space.
Fourier or Burg transforming this pattern into z-space generates a peak at the thickness of the irradiated sample.
It could be argued that though such a technique may work for single uniform slabs of dielectric material, it will give
results of little or no significance when the sample both scatters the incident radiation and gives erratic reflectivities due
to its non-uniform thickness and permittivity . We show results for a variety of materials such as explosive simulants,
powder and drugs, both alone and concealed under clothing or in a rucksack, which display strongly directional
reflectivities at millimeter wavelengths, and whose location is well displayed by a varying thickness parameter as the
millimetre beam is scanned across the target.
With this system we find that samples can easily be detected at standoff distances of at least 4.6m.
The charge-tube method is an accurate and efficient way of assigning the space-charge of a beam in computational simulations of charged particle systems. The method makes use of the trajectory steps that result from the process of trajectory integration. The space-charge associated with each step of each trajectory is assigned to a narrow cylindrical tube that surrounds the step. The total space-charge of a beam is then the sum of the charges in all the resulting the tubes. In systems of 2-dimensional axial symmetry the charge tubes become conical sheets of charge, and for some purposes these need to be given a finite thickness. The charge-tube method is particularly useful for simulating the space-charge of beams that are very narrow compared with their length. The implementation of the method is described and results obtained with it are compared with those obtained by the traditional charge-cell method.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.