The mission of the Department of Defense (DoD) Counter-drug Technology Development Program Office's Face Recognition Technology (FERET) program is to develop automatic face recognition systems from the development of detection and recognition algorithms in the laboratory through their demonstration in a prototype real-time system. To achieve this objective, the program supports research in face recognition algorithms, the collection of a large database of facial images, independent testing and evaluation of face recognition algorithms, construction of a real-time demonstration systems, and the integration of algorithms into the demonstration systems. The FERET program has established baseline performance for face recognition. The Army Research Laboratory (ARL) has been the technical agent for the Advanced Research Projects Agency since 1993, managing development of the recognition algorithms, database collection, and algorithm testing. Currently ARL is managing the development of several prototype face recognition systems that will demonstrate complete real-time video face identification in an access control mission. This paper gives an overview of the FERET program, presents recent performance results of face recognition algorithms evaluated, and addresses the future direction of the program and applications for DoD and law enforcement.
One of the most important technologies absent in traditional and emerging frontiers of computing is the management of visual information. Faces are accessible `windows' into the mechanisms that govern our emotional and social lives. The corresponding face recognition tasks considered herein include: (1) Surveillance, (2) CBIR, and (3) CBIR subject to correct ID (`match') displaying specific facial landmarks such as wearing glasses. We developed robust matching (`classification') and retrieval schemes based on hybrid classifiers and showed their feasibility using the FERET database. The hybrid classifier architecture consist of an ensemble of connectionist networks--radial basis functions-- and decision trees. The specific characteristics of our hybrid architecture include (a) query by consensus as provided by ensembles of networks for coping with the inherent variability of the image formation and data acquisition process, and (b) flexible and adaptive thresholds as opposed to ad hoc and hard thresholds. Experimental results, proving the feasibility of our approach, yield (i) 96% accuracy, using cross validation (CV), for surveillance on a data base consisting of 904 images (ii) 97% accuracy for CBIR tasks, on a database of 1084 images, and (iii) 93% accuracy, using CV, for CBIR subject to correct ID match tasks on a data base of 200 images.
Recent advances in information technology offer excellent potential for law enforcement personnel to enhance their surveillance and assessment capabilities. These personnel, regardless of their location, will be able to rapidly access critical information both from the scene of an incident and from archived records in a central database. Gathering information quickly from remote locations enables decision makers to accurately assess the seriousness of an incident, and in turn, provide the responders at the scene with the supporting information they need to do their jobs safely and effectively.
This paper presents initial results in a study comparing the effectiveness of visible and infra-red (IR) imagery for detecting and recognizing faces in areas where personnel identification is critical, (e.g., airports and secure buildings). We compare the effectiveness of visible versus IR imagery by running three face recognition algorithms on a database of images collected for this study. There are both IR and visible images for each person in the database collected using the same scenarios. We used three very different feature-extraction and decision-making algorithms for our study to insure that the comparisons would not depend on a particular processing technique. We also present recognition results when visible and infra-red decision metrics are fused. The recognition results show that both visible and IR imagery perform similarly across algorithms and that fusion of IR and visible imagery is a viable means of enhancing performance beyond that of either acting alone. We examine the relative importance of different regions of the face for recognition. We also discuss practical issues of implementation, along with plans for the next phase of the study, face detection in an uncontrolled environment. Preliminary face detection experiments are described.
This paper describes an approach for singularity (cores and deltas) detection on fingerprint images which uses fuzzy concepts and hierarchical structure. Singularity detection is usually performed of a discrete orientation matrix. Each of the elements of this matrix represents the dominant orientation on a given region. In this case, a fuzzy description of the orientation is defined. A very common method for singularity detection is the evaluation of the Poincare index on the orientation matrix. The fuzzy orientation is then naturally extended to the Poincare index. A multi-resolution description of the orientation matrix is developed and applied in a hierarchical way to estimate the location of singularities in fingerprint images. The method is illustrated on a database consisting of 310 fingerprints with mixed quality with application to fingerprint verification. The performance of this procedure is better than 98%.
Successful intrusion detection and assessment within a secured area typically requires the presentation of a large amount of information to a central alarm station operator. Typically, alarm information from several separate sensors is forwarded to the operator to be manually interpreted. However, the information is not always consistent with the actual situation and may not be sufficiently complete to make an accurate analysis. Alarm processing and fusion techniques can be applied to the sensor data to result in complete and manageable information that is presented to the operator for easier interpretation. This paper discusses the sensor fusion approach taken to combine the information from the three sensors included in the Advanced Exterior Sensor (AES). The AES is an intrusion detection and assessment system designed for wide-area coverage, quick deployment, low false/nuisance alarm operation, and immediate visual assessment. It combines three sensor technologies, visible, infrared, and millimeter wave radar, collocated on a compact and portable remote sensor module. The remote sensor module rotates at a rate of one revolution per second to detect and track motion and provide assessment in a continuous 360 degree(s) field-of-regard. Sensor fusion techniques are used to correlate and integrate the track data from the three sensors into a single track for operator observation. Additional inputs to the fusion process include environmental data, knowledge of sensor performance under certain weather conditions, sensor priority, and recent operator feedback. A confidence value is assigned to the track as a result of the fusion process. This helps to reduce nuisance alarms and to increase operator confidence in the system while reducing the workload of the operator.
This paper addresses the problem of improving detection, assessment, and response capabilities of security systems. Our approach combines two state-of-the-art technologies: volumetric video motion detection (VVMD) and virtual reality (VR). This work capitalizes on the ability of VVMD technology to provide 3D information about the position, shape, and size of intruders within a protected volume. The 3D information is obtained by fusing motion detection data from multiple video sensors. Other benefits include low nuisance alarm rates, increased resistance to tampering, low-bandwidth requirements for sending detection data to a remote monitoring site, and the ability to perform well in a dynamic environment where human activity and motion clutter are commonplace. The second component of this work involves the application of VR technology to display information relating to the sensors and the sensor environment. VR technology enables an operator, or security guard, to be immersed in a 3D graphical representation of the remote site containing the video sensors. VVMD data is transmitted from the remote site via ordinary telephone lines and displayed in real-time within the virtual environment. There are several benefits to displaying VVMD information in this way. Often, raw sensor information is not in a form that can be easily interpreted and understood--especially when taken out of the context of the sensor environment. Because the VVMD system provides 3D information and because the sensor environment is a physical 3D space, it seems natural to display this information in 3D. Also, the 3D graphical representation depicts essential details within and around the protected volume in a natural way for human perception. Sensor information can also be more easily interpreted when the operator can `move' through the virtual environment and explore the relationships between the sensor data, objects and other visual cues present in the visual environment. By exploiting the powerful ability of humans to understand and interpret 3D information, we expect to (1) improve the means for visualizing and interpreting sensor information, (2) allow a human operator to assess a potential threat more quickly and accurately, and (3) enable a more effective response. This paper will detail both the VVMD and VR technologies and will discuss a prototype system based upon their integration.
A short duration evaluation of a variety of commercially available Video Motion Detection products served to demonstrate application in detecting intrusion into protected areas of a military flightline. To be effective on a military flightline requires coverage over wide areas, adaptability to thermal imagery and ease of operator use. Opportunities exist to incorporate panning cameras with video motion detection to expand the awareness of Air Force Security Police. The interest in maximizing the utilization of relatively expensive and highly capable thermal imager cameras makes this a high interest priority. In the evaluation, the available products demonstrated an ability to perform intrusion detection, but flexibility in set-up and operation were seen as areas where improvement will be welcome.
In todays world of rapid technology development, emerging technologies present to law enforcement agencies, one of a very limited number of legal responses to criminal initiatives. Yet in spite of the existence of very sophisticated enforcement technologies, many law enforcement organizations seem to lag behind in several important areas of technology response and adaptation. This paper will propose the idea that what is needed is a closer partnership between private industry`technology adaptors' and government officials who are the problem `identifiers'. This marriage of necessity would produce immediate operational results focussed on todays enforcement problems and tomorrows emerging solutions. The paper will also present the concept of the establishment of an independent body tasked with the creation of an enforcement technology `adaptor' registry for authorized and innovative, enforcement technology applications specialists. This registry would be created to assist enforcement agencies with the location and/or creation of new law enforcement technologies/products. These persons would locate and/or create quick response technology innovations developed to provide operational `fixes' and technologies designed to address emerging enforcement problems. The presentation will conclude by providing examples of up to 12 different existing technologies devices currently available (but not generally known by government agencies).
The possession of a weapon in the courtroom of our American justice system is a Federal offense. A weapon in the courtroom is a particularly serious problem for judges, plaintiffs, defendants, and bystanders because of the emotional nature and environment of the courtroom. Potential violators include organized criminals and gangsters, individuals with personal vendettas, and otherwise sane and reasonable people who, because of the emotion of the moment, lose sight of right and wrong in our justice system. Most Federal courthouses incorporate some sort of metal detection apparatus used to screen and prevent individuals from carrying personal weapons into courtrooms. These devices are considered conventional in the sense that they employ an active electromagnetic induction technique that is sensitive to the presence of metal. This technique provides no information concerning the massiveness and location of the offending metal object and is prone to false alarms. This work introduces an alternative method for weapons discrimination that is based on passive magnetics technology in concert with simple comparative algorithms such that massiveness and location of suspected weapons may be determined. Since this system provides information not available from conventional metal detectors, false alarms will be greatly reduced, allowing remote monitoring of all entrances into a courtroom from a single vantage point, thereby saving considerable resources used on personnel costs.
Military personnel, law-enforcement officers, and civilians face ever-increasing dangers from persons carrying concealed handguns and other weapons. In direct correspondence with this danger is a need for more sophisticated means of detecting concealed weapons. We have developed a novel concealed-weapons detector based on the principle of low- frequency magnetic imaging. The detector is configured as a portal, and constructs an image of electrically conductive objects transported through it with a potential spatial resolution of approximately 1 inch. Measurements on a breadboard version of the weapons detector have, to date, yielded a resolution of 2 inches. In operation, magnetic dipole radiation, emitted by transmitting antennas in the perimeter of the portal, is scattered from conductive objects and is picked up by receive antennas, also positioned around the portal. With sufficient measurements, each with a different geometry, a solution to the inverse scattering problem can be found. The result is an image of conductive objects in the detector. The detector is sensitive to all metals, semiconductors, and conductive composites. The measured conductivity image formed by the detector is combined with the video signal from a visible CCD camera to form a composite image of persons transiting the detector portal and the conductive objects they are carrying. Accompanying image recognition software could be used to determine the threat level of objects based upon shape, conductivity, and placement on the person of the carrier, and provide cueing, logging, or alarm functions to the operator if suspect weapons are identified. The low- power, low-frequency emissions from the detector are at levels considered safe to humans and medical implants..
Millimeter-wave holographic imaging techniques have recently been developed for personnel surveillance applications at airports and other high-security checkpoints. Millimeter- wave imaging is useful for this application since millimeter-waves easily pass through common clothing materials yet are reflected from the human body and any items concealed by clothing. This allows a high-resolution imaging system to form an image revealing items concealed on the person imaged. A prototype imaging system developed at Pacific Northwest National Laboratory uses a scanned linear array of millimeter-wave antennas to capture wideband millimeter-wave data in approximately one second. This data is then mathematically reconstructed to form a high- resolution 3D image of the person being scanned. Millimeter- wave imaging has been demonstrated to be effective for detecting concealed weapons on personnel. Another imaging technique which could be applied to the weapon detection problem is acoustic imaging. Like millimeter-waves, ultrasonic acoustic waves can also penetrate clothing, and can be used to form relatively high-resolution images which can reveal concealed weapons on personnel. Acoustic imaging results have been obtained using wideband holographic imaging techniques nearly identical to the imaging techniques used for millimeter-wave imaging. Preliminary imaging results at 50 kHz indicate that acoustic imaging can be used to penetrate some types of common clothing materials. Hard clothing materials, such as leather on vinyl, are essentially opaque to acoustic waves at 50 kHz. In this paper, millimeter-wave and acoustic wave imaging techniques are compared for their effectiveness and suitability in weapon detection imaging systems. Experimental results from both imaging modalities are shown.
The Enclosed Space Detection System is a device to detect the presence of persons hiding in the enclosed spaces of a vehicle. The system operates by detecting the presence of the human ballistocardiogram.
First generation electronic monitoring systems are being used by the criminal justice system to effect behavioral modifications of persons in pre-trial release programs, on parole, and on probation. Current systems are merely radio frequency proximity detection systems that operate over limited ranges, on the order of 45 to 70 meters. One major defect with proximity detection systems is that when the offenders leave the area being monitored, there is no way to ensure that the offenders travel where they should. As a result, the first generation electronic monitoring systems are only applied to a restricted number of low risk cases. There is a growing need for a second generation electronic monitoring system which utilizes community-wide tracking and location technologies to increase the public safety and to expand the number of offenders monitored by these systems. Even though GPS (Global Positioning System) is rapidly becoming the technology of choice for vehicle tracking and location, GPS is not an ideal candidate for the second generation electronic monitoring system. Urban environments prevent GPS systems from providing continuous and accurate location service due to satellite occlusion by obstacles such as: hills, mountains, vehicles, buildings, and trees. An inverse-GPS approach which overcomes these urban environment related limitations has been evaluated by Northrop Grumman as a means to track people. This paper presents the results of a National Institute of Justice funded program to demonstrate in downtown Pittsburgh the feasibility of spread spectrum based time-of-arrival location systems for intelligently tracking people on probation and parole.
Nighttime surveillance is a key task for all law enforcement organizations. Traditional light intensification night vision systems suffer from poor sensitivity in extremely low light level situations, and are prone to blooming if an unexpected bright source intrudes upon the field of view. Thermal imagers detect infrared radiation emitted by all objects in proportion to the target's temperature, and are effective even in total darkness. However, until recently, most commercially available systems required cryogenic cooling, resulting in long start up times as Stirling engines drove the infrared detectors down to operating temperatures. This operational delay, combined with the audible noise emitted by many cryogenically cooled systems, resulted in systems which were less than optimum for law enforcement applications. This paper will describe a new uncooled microbolometer based infrared camera suitable for surveillance requirements. Basic microbolometer focal plane technology will be reviewed, and a description of its implementation into a hand held uncooled thermal imaging systems will be presented.
SAF Security Police are seeking to extend the application of commercial off-the-shelf (COTS) thermal imaging systems to the assessment and detection ofthreats to DoD personnel and resources. Cost and performance benefits are the prime motivaters to the military user during a period in which advanced imaging technology is being relied upon to offset the drawdown of security forces. Low-cost thermal imagers characterized by thermo-electrically cooled or stabilized room temperature detector arrays are offering a cost competitive alternative to traditional visible light surveillance cameras. Air Force Material Command has been testing and fielding two COTS thermal imagers within the last two years and is now supporting wider DoD application and procurement ofthis new class ofthermal imagers. Recently, the Security Systems Product Group at Hanscom, AFB conducted evaluations, procurement, and field operations ofboth the Hughes AN/PAS-20 hand held thermal imager and the Texas Instruments "Nightsight" Thermal Viewer. The Security Police identified the requirement for highly portable thermal imagers as part ofthe rapidreaction Tactical Automated Security System (TASS). The TASS mission is geared to peacekeeping and wartime protection of personnel, facilities and aircraft worldwide. This nimble, flexible capability represents the current state of the art in providing force protection during worldwide projection of military forces. The TASS will typically be shipped and installed within 48 hours to potential trouble spots such as Somalia, Bosnia or South West Asia. The versatility of a hand held imager with multiple power sources enables the operator to immediately visually detect intruders or to monitor detection sensors installed as a temporary or semi-permanent security perimeter. The Hughes AN/PAS-20 Hand Held Thermal Imager was tested and is being procured to meet this mission.. Previous field testing of a variety of remotely controlled thermal imagers occurred in 1993 at Vandenberg AFB, CA. That field testing, although only representative ofthe broad selection ofthermal imaging cameras, did provide an operational baseline for the siting of imagers relative to their performance under degraded weather conditions. In addition, the Air Force Security Police user The Texas Instruments "Nightsight" has been mounted on a commercial and military security vehicle, an elevating tripod, a roof-top, a light pole or fixed post to provide a panoramic, remotely-controllable thermal imager for mounted response teams, short range detection, perimeter alarm monitoring, and electronic image-based detection. Both of these systems are considered "low-cost", medium performance thermal imagers when compared with higher performance, dual-FOV thermal imagers costing more than $35,000 each. Generally, a minimum imaging system performance expectation of .10 C. Minimum Resolvable Temperature Difference (MIRTD) will enable the operator to pick out threats or targets out of backgrounds at ranges typical of the TASS deployment scenarios. The Air Force's current use ofthese devices matches commercial security requirements for high value assets such as aircraft, power plants, manufacturing facilities, and transportation equipment as well as anti-terrorism protection for VIP's and the general public. Furthermore, the future DoD application of these imagers to automated detection and smart sensor suites drives the near term integration and acquisition of this class of imager. A major concern to the military user is determining the optimal test methods able to be used to qualify commercial systems for military use without inflating the cost of the system. Often the military acquisition planner must present trade-off analyses to the military user enabling a lower cost, dual use approach to military acquisition. The critical compromises are often linked to actual versus perceived threat and mission. This paper describes recent testing and operational experience ofthe Texas Instruments Nightsight vehicle mounted thermal imager and the Hughes ANIPAS-20 hand held thermal imager. In addition, we review some ofthe tactical and cost factors which can be used by commercial physical security operators when selecting a suitable security imaging system.
Adverse weather, lack of background illumination and interference from artificial light sources can considerably reduce the effectiveness of passive LLLTV systems. Viewing in such conditions can be considerably enhanced by the addition of a pulsed laser illuminator synchronized to a gated LLLTV. Since the system has its own light source, surveillance is possible in zero or ultra-low light conditions. Furthermore using the technique of range gating, the effects of backscattering that occur in poor visibility conditions can be greatly reduced. The high resolution and self-illumination capability of the system permits positive identification of targets that would be undetectable or unresolvable with FLIR type systems.
Uncooled infrared imaging technology provides a new, affordable, high performance tool for both conventional and emerging applications in the surveillance and law enforcement markets. This technology offers users significant advantages, including high reliability, low power consumption, excellent image clarity, and good performance in adverse environments. This paper briefly discusses the various technologies used for night imaging, the advantages that infrared systems offer law enforcement and surveillance applications, how uncooled infrared systems fit into this market, and the status of uncooled microbolometer systems at Lockheed Martin.
The nation's first operational trial of a gunshot location system is underway in Redwood City, California. The system uses acoustic sensors widely distributed over an impacted community. The impulses received at each sensor allow triangulation of the gunfire location and prompt police dispatch. A computer display shows gunfire location superimposed on a map showing property boundaries. Police are responding to system events immediately and in a community-policing investigative role.