This PDF file contains the front matter associated with SPIE Proceedings Volume 9995, including the Title Page, Copyright information, Table of Contents, Introduction (if any), and Conference Committee listing.

Security holograms having unique 3D images are one of the tools for enhancing the security for product and personnel authentication and anti-counterfeiting. Apart from the high technology that is required, the uniqueness of a 3D object presents a significant additional threshold for the counterfeiting of such security holograms. But, due to the development of 3D printing technology, the hurdles are disabled and allow the chances of counterfeiting. In order to overcome this, holographic interferometry is effectively utilized and the object is recorded twice before and after the state of random object change. At the time of reconstruction, two signal waves generated simultaneously interfere each other, resulting in a fringe modulation. This fringe modulation in 3D image hologram with respect to the random object change is exploited to generate a rigid and unique anticounterfeit feature. Though holographic interferometry techniques are being widely used for the non-destructive evaluation, the applicability of this technology for the security and forensic activity is less exploited. This paper describes our efforts to introduce holographic interferometry in 3D image holograms for security and forensic applications.

Authentication and optical verification of travel documents upon crossing borders is of utmost importance for national security. Understanding the workflow and different approaches to ICAO 9303 travel document scanning in passport readers, as well as highlighting normalization issues and designing new methods to achieve better harmonization across inspection devices are key steps for the development of more effective and efficient next- generation passport inspection. This paper presents a survey of state-of-the-art document inspection systems, showcasing results of a document reader challenge investigating 9 devices with regards to optical characteristics.

In this study, we report on the recent developments of optical image authentication algorithms. Compared with conventional optical encryption, optical image authentication achieves more security strength because such methods do not need to recover information of plaintext totally during the decryption period. Several recently proposed authentication systems are briefly introduced. We also propose a novel multiple-image authentication system, where multiple original images are encoded into a photon-limited encoded image by using a triple-plane based phase retrieval algorithm and photon counting imaging (PCI) technique. One can only recover a noise-like image using correct keys. To check authority of multiple images, a nonlinear fractional correlation is employed to recognize the original information hidden in the decrypted results. The proposal can be implemented optically using a cascaded phase filtering configuration. Computer simulation results are presented to evaluate the performance of this proposal and its effectiveness.

X-ray imaging is the most commonly used method in baggage screening. Conventional x-ray attenuation (usually in dual-energy mode) is exploited to discriminate threat and non-threat materials: this is essentially, a method that has seen little changes in decades. Our goal is to demonstrate that x-rays can be used in a different way to achieve improved detection of weapons and explosives. Our approach involves the use of x-ray phase contrast and it a) allows much higher sensitivity in the detection of object edges and b) can be made sensitive to the sample’s microstructure. We believe that these additional channels of information, alongside conventional attenuation which would still be available, have the potential to significantly increase both sensitivity and specificity in baggage scanning.

We obtained preliminary data demonstrating the above enhanced detection, and we built a scanner (currently in commissioning) to scale the concept up and test it on real baggage. In particular, while previous X-ray phase contrast imaging systems were limited in terms of both field of view (FOV) and maximum x-ray energy, this scanner overcomes both those limitations and provides FOVs up to 20 to 50 cm² with x-ray energies up to 100 keV.

Raman spectroscopy allows the acquisition of molecularly specific signatures of pure compounds and mixtures making it a popular method for material identification applications. In hazardous materials, security and counter terrorism applications, conventional handheld Raman systems are typically limited to operation by line-of-sight or through relatively transparent plastic bags / clear glass vials. If materials are concealed behind thicker, coloured or opaque barriers it can be necessary to open and take a sample. Spatially Offset Raman Spectroscopy (SORS)^[1] is a novel variant of Raman spectroscopy whereby multiple measurements at differing positions are used to separate the spectrum arising from the sub layers of a sample from the spectrum at the surface. For the first time, a handheld system based on SORS has been developed and applied to hazardous materials identification. The system - "Resolve" - enables new capabilities in the rapid identification of materials concealed by a wide variety of non-metallic sealed containers such as; coloured and opaque plastics, paper, card, sacks, fabric and glass. The range of potential target materials includes toxic industrial chemicals, explosives, narcotics, chemical warfare agents and biological materials. Resolve has the potential to improve the safety, efficiency and critical decision making in incident management, search operations, policing and ports and border operations. The operator is able to obtain a positive identification of a potentially hazardous material without opening or disturbing the container - to gain access to take a sample - thus improving safety. The technique is fast and simple thus suit and breathing gear time is used more efficiently. SORS also allows Raman to be deployed at an earlier stage in an event before more intrusive techniques are used. Evidential information is preserved and the chain of custody protected. Examples of detection capability for a number of materials and barrier types are presented below.

In order to counter drug-related crimes effectively, and to safeguard homeland security as well as public safety, it is important to inspect drugs, explosives and other contraband quickly and accurately from the express mail system, luggage, vehicles and other objects.

In this paper, we discuss X-ray backscatter inspection system based on a novel lobster-eye X-ray objective, which is an effective inspection technology for drugs, explosives and other contraband inspection. Low atomic number materials, such as drugs and explosives, leads to strong Compton scattering after irradiated by X-ray, which is much stronger than high atomic number material, such as common metals, etc. By detecting the intensity of scattering signals, it is possible to distinguish between organics and inorganics. The lobster-eye X-ray optical system imitates the reflective eyes of lobsters, which field of view can be made as large as desired and it is practical to achieve spatial resolution of several millimeters for finite distance detection. A novel lobster-eye X-ray objective is designed based on modifying Schmidt geometry by using multi-lens structure, so as to reduce the difference of resolution between the horizontal and vertical directions. The demonstration experiments of X-ray backscattering imaging were carried out. A suitcase, a wooden box and a tire with several typical samples hidden in them were imaged by the X-ray backscattering inspection system based on a lobster-eye X-ray objective. The results show that this X-ray backscattering inspection system can get a resolution of less than five millimeters under the FOV of more than two hundred millimeters with ~0.5 meter object distance, which can still be improved.

The spectrum of mid-infrared light scattered from an actively illuminated aerosol was used to distinguish between different chemicals. Using spectrally broad illumination from an optical parametric oscillator covering 3.2 – 3.55 μm, characteristic absorption features of two different chemicals were detected, and two similar molecules were clearly distinguished using the spectra of backscattered light from each chemical aerosol.

The development of portable sensors for fast screening of crime scenes is required to reduce the number of evidences useful to be collected, optimizing time and resources. Laser based spectroscopic techniques are good candidates to this scope due to their capability to operate in field, in remote and rapid way.

In this work, the prototype of a multispectral imaging LIF (Laser Induced Fluorescence) system able to detect evidence of different materials on large very crowded and confusing areas at distances up to some tens of meters will be presented. Data collected as both 2D fluorescence images and LIF spectra are suitable to the identification and the localization of the materials of interest. A reduced scan time, preserving at the same time the accuracy of the results, has been taken into account as a main requirement in the system design. An excimer laser with high energy and repetition rate coupled to a gated high sensitivity ICCD assures very good performances for this purpose. Effort has been devoted to speed up the data processing. The system has been tested in outdoor and indoor real scenarios and some results will be reported. Evidence of the plastics polypropylene (PP) and polyethilene (PE) and polyester have been identified and their localization on the examined scenes has been highlighted through the data processing. By suitable emission bands, the instrument can be used for the rapid detection of other material classes (i.e. textiles, woods, varnishes).

The activities of this work have been supported by the EU-FP7 FORLAB project (Forensic Laboratory for in-situ evidence analysis in a post blast scenario).

Biological hazardous substances like certain fungi and bacteria represent a high risk for the broad public if fallen into wrong hands. Incidents based on bio agents are commonly considered to have incalculable and complex consequences for first responders and people. The impact of such an event can be minimized by a combination of different sensor technologies that have been developed to detect bio-threats and to gather information after an incident. Sensors for bio-agents can be grouped into two categories. Sampling devices collect material from locations supposed to be contaminated, and they are able to identify biological material with high sensitivity and selectivity. However, these point sensors need to be positioned correctly in advance of an attack, and moving sources of biological material cannot be tracked. A different approach is based on optical standoff detection. For biological samples laser induced florescence (LIF) has been proven to get real time data on location and type of hazards without being in contact with the suspicious substance. This work is based on a bio-detector developed at the DLR Lampoldshausen. The LIF detection has been designed for outdoor operation at standoff distances from 20 m up to more than 100 m. The detector acquires LIF spectral data for two different excitation wavelengths (280 and 355 nm) as well as time resolved information for the fluorescence decay which can be used to classify suspicious samples. While the classification device had been trained on uncritical samples (like amino acids, NADH, yeast, chemicals, oils), this work presents the progress to more relevant, living bacteria of different strains. The low risk and non-pathogenic bacteria Bacillus thuringensis, Bacillus atrophaeus, Bacillus subtilis, Brevibacillus brevis, Micrococcus luteus, Oligella urethralis, Paenibacillus polymyxa and Escherichia coli (K12) have been investigated with the above set-up at both excitation wavelengths

Detection, Identification and Monitoring (DIM) of biological material is critical to enhancing Situational Awareness (SA) in a timely manner, supporting decisions, and enabling the endangered force to take the most appropriate actions in a recognized CB environment. An optimum Bio DIM capability would include both point sensors to provide local monitoring and sampling for confirmatory ID of the material, and standoff sensors to provide wide-area monitoring from a distance, increasing available response time and enhancing SA. In June 2015, a Canadian team co-deployed a point (VPBio) and a standoff (BioSense) bio sensor during the international S/K Challenge II event, at Dugway Proving Ground (DPG), USA. The co-deployment of the point and standoff sensors allowed the assessment of their respective strengths and limitations with regards to Bio DIM and SA in a realistic CB environment. Moreover, the initial hypothesis stating the existence of valuable leverages between the two sensors in a context of Bio DIM was confirmed. Indeed, the spatial limitation of the point sensor was overcome with the wide area coverage of the standoff technology. In contrast, the sampling capability of the point sensor can allow confirmatory identification of the detected material. Additionally, in most scenarios, the combined results allowed an increase in detection confidence. In conclusion, the demonstration of valuable leverages between point and standoff sensors in a context of Bio DIM was made, confirming the mitigation effect of co-deploying these systems for bio surveillance.

Aviation security requirements adopted in 2014 require liquids to be screened at most airports throughout Europe, North America and Australia. Cobalt’s unique Spatially Offset Raman Spectroscopy (SORS™) technology has proven extremely effective at screening liquids, aerosols and gels (LAGS) with extremely low false alarm rates. SORS is compatible with a wide range of containers, including coloured, opaque or clear plastics, glass and paper, as well as duty-free bottles in STEBs (secure tamper-evident bags). Our award-winning Insight range has been specially developed for table-top screening at security checkpoints. Insight systems use our patented SORS technology for rapid and accurate chemical analysis of substances in unopened non-metallic containers. Insight100M™ and the latest member of the range - Insight200M™ - also screen metallic containers. Our unique systems screen liquids, aerosols and gels with the highest detection capability and lowest false alarm rates of any ECAC-approved scanner, with several hundred units already in use at airports including eight of the top ten European hubs. This paper presents an analysis of real performance data for these systems.

Raman spectroscopy is a well-established spectroscopic method for the detection of condensed phase chemicals. It is based on scattered light from exposure of a target material to a narrowband laser beam. The information generated enables presumptive identification from measuring correlation with library spectra. Whilst this approach is successful in identification of chemical information of samples with one component, it is more difficult to apply to spectral mixtures. The capability of handling spectral mixtures is crucial for defence and security applications as hazardous materials may be present as mixtures due to the presence of degradation, interferents or precursors. A novel method for spectral unmixing is proposed here. Most modern decomposition techniques are based on the sparse decomposition of mixture and the application of extra constraints to preserve the sum of concentrations. These methods have often been proposed for passive spectroscopy, where spectral baseline correction is not required. Most successful methods are computationally expensive, e.g. convex optimisation and Bayesian approaches.

We present a novel low complexity sparsity based method to decompose the spectra using a reference library of spectra. It can be implemented on a hand-held spectrometer in near to real-time. The algorithm is based on iteratively subtracting the contribution of selected spectra and updating the contribution of each spectrum. The core algorithm is called fast non-negative orthogonal matching pursuit, which has been proposed by the authors in the context of nonnegative sparse representations. The iteration terminates when the maximum number of expected chemicals has been found or the residual spectrum has a negligible energy, i.e. in the order of the noise level. A backtracking step removes the least contributing spectrum from the list of detected chemicals and reports it as an alternative component. This feature is particularly useful in detection of chemicals with small contributions, which are normally not detected. The proposed algorithm is easily reconfigurable to include new library entries and optional preferential threat searches in the presence of predetermined threat indicators.

Under Ministry of Defence funding, we have demonstrated the algorithm for fingerprinting and rough quantification of the concentration of chemical mixtures using a set of reference spectral mixtures. In our experiments, the algorithm successfully managed to detect the chemicals with concentrations below 10 percent. The running time of the algorithm is in the order of one second, using a single core of a desktop computer.

The requirement for a modern method of the detection of explosives is still important. One of the promising areas is laser remote detection of explosives. In this work we have developed a portable laser and quantitatively investigated its use for sampling of common explosives on different surfaces. It has been shown that most of desorbed products consist of molecules of explosives themselves, and the proportion of dissociated fragments is small.

It is shown that laser irradiation increases the amount of desorbed matter not only by increasing the number of pulses, but also due to an additional warming of a substrate. The use of frequency mode smoothens peculiar properties of the desorption process for all explosives and substrates used. The typical desorbed amount ranges from 95 ng to 7900 ng for 10-second irradiation of the samples used.

The developed laser can be applied in ion mobility spectrometry, IMS-based detectors of explosives to increase the distance of sampling. It has a mass of less than 2kg.

We discuss the possibility of the detection and identification of substances under real conditions. As it is well-known many harmless materials show similar spectral properties to hazardous substances. To identify a matter we are currently developing various time-dependent criteria for the estimation of the probability of a hazardous substance presence in a sample under analysis. A new type of such criteria is discussed in this paper. To explain a physical mechanism for false alarm spectral properties appearance in the signal transmitted through and reflected from a substance we make a computer simulation using 1D Maxwell's equations and density matrix formalism. A feasible approach for highly effective detection and identification of a substance is proposed.

This paper is a continuation of the work of Becker et al.¹ In their work, they analyzed the robustness of various background subtraction algorithms on fused video streams originating from visible and infrared cameras. In order to cover a broader range of background subtraction applications, we show the effects of fusing infrared-visible video streams from vibrating cameras on a large set of background subtraction algorithms. The effectiveness is quantitatively analyzed on recorded data of a typical outdoor sequence with a fine-grained and accurate annotation of the images. Thereby, we identify approaches which can benefit from fused sensor signals with camera jitter. Finally conclusions on what fusion strategies should be preferred under such conditions are given.

Wide Area Motion Imagery (WAMI) enables image based surveillance of areas that can cover multiple square kilometers. Interpreting and analyzing information from such sources, becomes increasingly time consuming as more data is added from newly developed methods for information extraction. Captured from a moving Unmanned Aerial Vehicle (UAV), the high-resolution images allow detection and tracking of moving vehicles, but this is a highly challenging task. By using a chain of computer vision detectors and machine learning techniques, we are capable of producing high quality track information of more than 40 thousand vehicles per five minutes. When faced with such a vast number of vehicular tracks, it is useful for analysts to be able to quickly query information based on region of interest, color, maneuvers or other high-level types of information, to gain insight and find relevant activities in the flood of information.

In this paper we propose a set of tools, combined in a graphical user interface, which allows data analysts to survey vehicles in a large observed area. In order to retrieve (parts of) images from the high-resolution data, we developed a multi-scale tile-based video file format that allows to quickly obtain only a part, or a sub-sampling of the original high resolution image. By storing tiles of a still image according to a predefined order, we can quickly retrieve a particular region of the image at any relevant scale, by skipping to the correct frames and reconstructing the image. Location based queries allow a user to select tracks around a particular region of interest such as landmark, building or street. By using an integrated search engine, users can quickly select tracks that are in the vicinity of locations of interest. Another time-reducing method when searching for a particular vehicle, is to filter on color or color intensity. Automatic maneuver detection adds information to the tracks that can be used to find vehicles based on their behavior.

Object recognition is important to understand the content of video and allow flexible querying in a large number of cameras, especially for security applications. Recent benchmarks show that deep convolutional neural networks are excellent approaches for object recognition. This paper describes an approach of domain transfer, where features learned from a large annotated dataset are transferred to a target domain where less annotated examples are available as is typical for the security and defense domain. Many of these networks trained on natural images appear to learn features similar to Gabor filters and color blobs in the first layer. These first-layer features appear to be generic for many datasets and tasks while the last layer is specific. In this paper, we study the effect of copying all layers and fine-tuning a variable number. We performed an experiment with a Caffe-based network on 1000 ImageNet classes that are randomly divided in two equal subgroups for the transfer from one to the other. We copy all layers and vary the number of layers that is fine-tuned and the size of the target dataset. We performed additional experiments with the Keras platform on CIFAR-10 dataset to validate general applicability. We show with both platforms and both datasets that the accuracy on the target dataset improves when more target data is used. When the target dataset is large, it is beneficial to freeze only a few layers. For a large target dataset, the network without transfer learning performs better than the transfer network, especially if many layers are frozen. When the target dataset is small, it is beneficial to transfer (and freeze) many layers. For a small target dataset, the transfer network boosts generalization and it performs much better than the network without transfer learning. Learning time can be reduced by freezing many layers in a network.

Existing tracking methods vary strongly in their approach and therefore have different strengths and weaknesses. For example, a single tracking algorithm may be good at handling variations in illumination, but does not cope well with deformation. Hence, their failures can occur on entirely different time intervals on the same sequence. One possible solution for overcoming limitations of a single tracker and for benefitting from individual strengths, is to run a set of tracking algorithms in parallel and fuse their outputs. But in general, tracking algorithms are not designed to receive feedback from a higher level fusion strategy or require a high degree of integration between individual levels. Towards this end, we introduce a fusion strategy serving the purpose of online single object tracking, for which no knowledge about individual tracker characteristics is needed. The key idea is to combine several independent and heterogeneous tracking approaches and to robustly identify an outlier subset based on the "Median Absolute Deviations" (MAD) measure. The MAD fusion strategy is very generic and only requires frame-based object bounding boxes as input. Thus, it can work with arbitrary tracking algorithms. Furthermore, the MAD fusion strategy can also be applied for combining several instances of the same tracker to form a more robust ensemble for tracking an object. The evaluation is done on public available datasets. With a set of heterogeneous, commonly used trackers we show that the proposed MAD fusion strategy improves the tracking results in comparison to a classical combination of parallel trackers and that the tracker ensemble helps to deal with the initialization uncertainty of a single tracker.

Person re-identification is the problem of matching multiple occurrences of a person in large amounts of image or video data. In this work we propose an approach specifically tailored to re-identify people across different camera views in aerial video recordings. New challenges that arise in aerial data include unusual and more varied view angles, a moving camera and potentially large changes in environment and other in uences between recordings (i.e. between flights). Our approach addresses these new challenges. Due to their recent successes, we apply deep learning to automatically learn features for person re-identification on a number of public datasets. We evaluate these features on aerial data and propose a method to automatically select suitable pretrained features without requiring person id labels on the aerial data. We further show that tailored data augmentation methods are well suited to better cope with the larger variety in view angles. Finally, we combine our model with a metric learning approach to allow for interactive improvement of re-identification results through user feedback. We evaluate the approach on our own video dataset which contains 12 persons recorded from a UAV.

Deception detection is valuable in the security domain to distinguish truth from lies. It is desirable in many security applications, such as suspect and witness interviews and airport passenger screening. Interviewers are constantly trying to assess the credibility of a statement, usually based on intuition without objective technical support. However, psychological research has shown that humans can hardly perform better than random guessing. Deception detection is a multi-disciplinary research area with an interest from different fields, such as psychology and computer science. In the last decade, several developments have helped to improve the accuracy of lie detection (e.g., with a concealed information test, increasing the cognitive load, or measurements with motion capture suits) and relevant cues have been discovered (e.g., eye blinking or fiddling with the fingers). With an increasing presence of mobile phones and bodycams in society, a mobile, stand-off, automatic deception detection methodology based on various cues from the whole body would create new application opportunities. In this paper, we study the feasibility of measuring these visual cues automatically on different parts of the body, laying the groundwork for stand-off deception detection in more flexible and mobile deployable sensors, such as body-worn cameras. We give an extensive overview of recent developments in two communities: in the behavioral-science community the developments that improve deception detection with a special attention to the observed relevant non-verbal cues, and in the computer-vision community the recent methods that are able to measure these cues. The cues are extracted from several body parts: the eyes, the mouth, the head and the fullbody pose. We performed an experiment using several state-of-the-art video-content-analysis (VCA) techniques to assess the quality of robustly measuring these visual cues.

A novel approach for autonomous detection of anomalies in crowded environments is presented in this paper. The proposed models uses a Gaussian mixture probability hypothesis density (GM-PHD) filter as feature extractor in conjunction with different Gaussian mixture hidden Markov models (GM-HMMs). Results, based on both simulated and recorded data, indicate that this method can track and detect anomalies on-line in individual crowds through multiple camera feeds in a crowded environment.

We are living in a world dependent on sophisticated technical infrastructure. Malicious manipulation of such critical infrastructure poses an enormous threat for all its users. Thus, running a critical infrastructure needs special attention to log the planned maintenance or to detect suspicious events. Towards this end, we present a knowledge-based surveillance approach capable of logging visual observable events in such an environment. The video surveillance modules are based on appearance-based person detection, which further is used to modulate the outcome of generic processing steps such as change detection or skin detection. A relation between the expected scene behavior and the underlying basic video surveillance modules is established. It will be shown that the combination already provides sufficient expressiveness to describe various everyday situations in indoor video surveillance. The whole approach is qualitatively and quantitatively evaluated on a prototypical scenario in a server room.

The bottleneck in situation awareness is no longer in the sensing domain but rather in the data interpretation domain, since the number of sensors is rapidly increasing and it is not affordable to increase human data-analysis capacity at the same rate. Automatic image analysis can assist a human analyst by alerting when an event of interest occurs. However, common state-of-the-art image recognition systems learn representations in high-dimensional feature spaces, which makes them less suitable to generate a user-comprehensive message. Such data-driven approaches rely on large amounts of training data, which is often not available for quite rare but high-impact incidents in the security domain. The key contribution of this paper is that we present a novel real-time system for image understanding based on generic instantaneous low-level processing components (symbols) and flexible user-definable and user-understandable combinations of these components (sentences) at a higher level for the recognition of specific relevant events in the security domain. We show that the detection of an event of interest can be enhanced by utilizing recognition of multiple short-term preparatory actions.

The majority of sensor installations for area protection (e.g. critical national infrastructure, military forward operating bases, etc.) make use of banks of screens each containing one or more sensor feeds, such that the burden of combining data from the various sources, understanding the situation, and controlling the sensors all lies with the human operator. Any automation in the system is generally heavily bespoke for the particular installation, leading to an inflexible system which is difficult to change or upgrade. We have developed a modular system architecture consisting of intelligent autonomous sensor modules, a high level decision making module, a middleware integration layer and an end-user GUI. The modules are all effectively "plug and play", and we have demonstrated that it is relatively simple to incorporate legacy sensors into the architecture. We have extended our previously-reported SAPIENT demonstration system to operate with a larger number and variety of sensor modules, over an extended area, detecting and classifying a wider variety of "threat activities", both vehicular and pedestrian. We report the results of a demonstration of the SAPIENT system containing multiple autonomous sensor modules with a range of modalities including laser scanners, radar, TI, EO, acoustic and seismic sensors. They operate from a combination of mains, generator and battery power, and communicate with the central “hub” over Ethernet, point-to-point wireless links and Wi-Fi. The system has been configured to protect an extended area in a complex semi-urban environment. We discuss the operation of the SAPIENT system in a realistic demonstration environment (which included significant activity not under trial control), showing sensor cueing, multi-modal sensor fusion, threat prioritisation and target hand-off.

SLATE is an Autonomous Sensor Module (ASM) designed to work with the SAPIENT system providing accurate location tracking and classifications of targets that pass through its field of view. The concept behind the SLATE ASM is to produce a sensor module that provides a complementary view of the world to the camera-based systems that are usually used for wide area surveillance. Cameras provide a hi-fidelity, human understandable view of the world with which tracking and identification algorithms can be used. Unfortunately, positioning and tracking in a 3D environment is difficult to implement robustly, making location-based threat assessment challenging. SLATE uses a Scanning Laser Rangefinder (SLR) that provides precise (<1cm) positions, sizes, shapes and velocities of targets within its field-of-view (FoV). In this paper we will discuss the development of the SLATE ASM including the techniques used to track and classify detections that move through the field of view of the sensor providing the accurate tracking information to the SAPIENT system. SLATE’s ability to locate targets precisely allows subtle boundary-crossing judgements, e.g. on which side of a chain-link fence a target is. SLATE’s ability to track targets in 3D throughout its FoV enables behavior classification such as running and walking which can provide an indication of intent and help reduce false alarm rates.

Real-time classification of vehicles into sub-category types poses a significant challenge within infra-red imagery due to the high levels of intra-class variation in thermal vehicle signatures caused by aspects of design, current operating duration and ambient thermal conditions. Despite these challenges, infra-red sensing offers significant generalized target object detection advantages in terms of all-weather operation and invariance to visual camouflage techniques. This work investigates the accuracy of a number of real-time object classification approaches for this task within the wider context of an existing initial object detection and tracking framework. Specifically we evaluate the use of traditional feature-driven bag of visual words and histogram of oriented gradient classification approaches against modern convolutional neural network architectures. Furthermore, we use classical photogrammetry, within the context of current target detection and classification techniques, as a means of approximating 3D target position within the scene based on this vehicle type classification. Based on photogrammetric estimation of target position, we then illustrate the use of regular Kalman filter based tracking operating on actual 3D vehicle trajectories. Results are presented using a conventional thermal-band infra-red (IR) sensor arrangement where targets are tracked over a range of evaluation scenarios.

Most surveillance systems combine camera sensors with other detection sensors that trigger an alert to a human operator when an object is detected. The detection sensors typically require careful installation and configuration for each application and there is a significant burden on the operator to react to each alert by viewing camera video feeds. A demonstration system known as Sensing for Asset Protection with Integrated Electronic Networked Technology (SAPIENT) has been developed to address these issues using Autonomous Sensor Modules (ASM) and a central High Level Decision Making Module (HLDMM) that can fuse the detections from multiple sensors.

This paper describes the 24 GHz radar based ASM, which provides an all-weather, low power and license exempt solution to the problem of wide area surveillance. The radar module autonomously configures itself in response to tasks provided by the HLDMM, steering the transmit beam and setting range resolution and power levels for optimum performance. The results show the detection and classification performance for pedestrians and vehicles in an area of interest, which can be modified by the HLDMM without physical adjustment. The module uses range-Doppler processing for reliable detection of moving objects and combines Radar Cross Section and micro-Doppler characteristics for object classification. Objects are classified as pedestrian or vehicle, with vehicle sub classes based on size. Detections are reported only if the object is detected in a task coverage area and it is classified as an object of interest.

The system was shown in a perimeter protection scenario using multiple radar ASMs, laser scanners, thermal cameras and visible band cameras. This combination of sensors enabled the HLDMM to generate reliable alerts with improved discrimination of objects and behaviours of interest.

A UK MoD funded programme into autonomous sensors arrays (SAPIENT) has been developing new, highly capable sensor modules together with a scalable modular architecture for control and communication. As part of this system there is a desire to also utilise existing legacy sensors. The paper reports upon the development of a SAPIENT-compliant sensor module using a legacy Close-Circuit Television (CCTV) pan-tilt-zoom (PTZ) camera. The PTZ camera sensor provides three modes of operation. In the first mode, the camera is automatically slewed to acquire imagery of a specified scene area, e.g. to provide “eyes-on” confirmation for a human operator or for forensic purposes. In the second mode, the camera is directed to monitor an area of interest, with zoom level automatically optimized for human detection at the appropriate range. Open source algorithms (using OpenCV) are used to automatically detect pedestrians; their real world positions are estimated and communicated back to the SAPIENT central fusion system. In the third mode of operation a “follow” mode is implemented where the camera maintains the detected person within the camera field-of-view without requiring an end-user to directly control the camera with a joystick.

Many existing asset/area protection systems, for example those deployed to protect critical national infrastructure, are comprised of multiple sensors such as EO/IR, radar, and Perimeter Intrusion Detection Systems (PIDS), loosely integrated with a central Command and Control (C2) system. Whilst some sensors provide automatic event detection and C2 systems commonly provide rudimentary multi-sensor rule based alerting, the performance of such systems is limited by the lack of deep integration and autonomy. As a result, these systems have a high degree of operator burden.

To address these challenges, an architectural concept termed "SAPIENT" was conceived. SAPIENT is based on multiple Autonomous Sensor Modules (ASMs) connected to a High-Level Decision Making Module (HLDMM) that provides data fusion, situational awareness, alerting, and sensor management capability. The aim of the SAPIENT concept is to allow for the creation of a surveillance system, in a modular plug-and-play manner, that provides high levels of autonomy, threat detection performance, and reduced operator burden.

This paper considers the challenges associated with developing an HLDMM aligned with the SAPIENT concept, through the discussion of the design of a realised HLDMM. Particular focus is drawn to how high levels of system level performance can be achieved whilst retaining modularity and flexibility. A number of key aspects of our HLDMM are presented, including an integrated threat assessment and sensor management framework, threat sequence matching, and ASM trust modelling. The results of real-world testing of the HLDMM, in conjunction with multiple Laser, Radar, and EO/IR sensors, in representative semi-urban environments, are discussed.

Two of the major new concerns of modern societies are biosecurity and biosafety. Several biological agents (BAs) such as toxins, bacteria, viruses, fungi and parasites are able to cause damage to living systems either humans, animals or plants. Optical techniques, in particular LIght Detection And Ranging (LIDAR), based on the transmission of laser pulses and analysis of the return signals, can be successfully applied to monitoring the release of biological agents into the atmosphere.

It is well known that most of biological agents tend to emit specific fluorescence spectra, which in principle allow their detection and identification, if excited by light of the appropriate wavelength. For these reasons, the detection of the UVLight Induced Fluorescence (UV-LIF) emitted by BAs is particularly promising. On the other hand, the stand-off detection of BAs poses a series of challenging issues; one of the most severe is the automatic discrimination between various agents which emit very similar fluorescence spectra.

In this paper, a new data analysis method, based on a combination of advanced filtering techniques and Support Vector Machines, is described. The proposed approach covers all the aspects of the data analysis process, from filtering and denoising to automatic recognition of the agents. A systematic series of numerical tests has been performed to assess the potential and limits of the proposed methodology. The first investigations of experimental data have already given very encouraging results.

Dangerous materials inspection is an important technique to confirm dangerous materials crimes. It has significant impact on the prohibition of dangerous materials-related crimes and the spread of dangerous materials. Lobster-Eye Optical Imaging System is a kind of dangerous materials detection device which mainly takes advantage of backscatter X-ray. The strength of the system is its applicability to access only one side of an object, and to detect dangerous materials without disturbing the surroundings of the target material.

The device uses Compton scattered x-rays to create computerized outlines of suspected objects during security detection process. Due to the grid structure of the bionic object glass, which imitate the eye of a lobster, grids contribute to the main image noise during the imaging process. At the same time, when used to inspect structured or dense materials, the image is plagued by superposition artifacts and limited by attenuation and noise. With the goal of achieving high quality images which could be used for dangerous materials detection and further analysis, we developed effective image process methods applied to the system. The first aspect of the image process is the denoising and enhancing edge contrast process, during the process, we apply deconvolution algorithm to remove the grids and other noises. After image processing, we achieve high signal-to-noise ratio image. The second part is to reconstruct image from low dose X-ray exposure condition. We developed a kind of interpolation method to achieve the goal. The last aspect is the region of interest (ROI) extraction process, which could be used to help identifying dangerous materials mixed with complex backgrounds. The methods demonstrated in the paper have the potential to improve the sensitivity and quality of x-ray backscatter system imaging.