Thrust 3
Explosives Detection Sensor Systems

The focus of this thrust is to develop the basic science to enable successful design and implementation of explosive detection and identification systems using multisensor systems and unconventional approaches including alternative signatures such as behavioral and motion cues. Of particular interest are hybrid systems that use multiple modalities that combine information about diverse and complementary properties of explosives, plus information about human factors. Measurable properties provided by the suite of sensors include x-ray absorption or backscatter characteristics at different energies, emission measurements in multiple spectral bands, geometric shape, nuclear composition, thermal and visual imagery as well as unconventional signatures like video motion patterns and anomalous behavior of suspected individuals. The basic science will focus on methodologies for selecting appropriate combinations of sensors, extracting critical features from each sensor, fusing the information across sensors for optimal detection and identification of threats, and actively controlling the sensing and processing of information to improve the sensitivity and specificity of explosive detection systems.

We plan to consider both portal and stand-off-detection systems, motivated by difficult explosives detection problems such as the detection of suicide bombers and vehicle-mounted and hand-carried IEDs. Portal detection systems offer the potential for controlled sensing environments and simultaneous acquisition of multisensor data with reduced clutter ambiguity. In contrast, stand-off detection systems work with weak signals in uncontrolled backgrounds, where simultaneous multisensor measurements and controlled sensing geometries are difficult to achieve. Stand-off multisensor systems will be based on networks of heterogeneous sensors fusing information collected over space and time. Our goal is to develop technology for systems that can scale to realistic throughputs, and hence will operate with significant amounts of autonomy, involving human operators primarily for critical decision tasks.

Project Leaders

  • W. Clem Karl
    Professor
    Boston University
    Email

  • Eric Miller
    Professor
    Tufts University
    Email

  • Charles A. Bouman
    Professor
    Purdue University
    Email

  • David Castañón
    Professor
    Boston University
    Email

  • Birsen Yazici
    Professor
    Rensselaer Polytechnic Institute
    Email

  • Octavia Camps
    Professor
    Northeastern University
    Email

  • Mario Sznaier
    Professor
    Northeastern University
    Email

  • Richard Radke
    Associate Professor
    Rensselaer Polytechnic Institute
    Email

  • Venkatesh Saligrama
    Associate Professor
    Boston University
    Email

  • Homer Pien
    Director, Laboratory for Medical Imaging & Computations
    Massachusetts General Hospital
    Email


Faculty and Staff

  • Ken D.Sauer
    Professor
    Notre Dame University
    Email

  • Gilead Tadmor
    Professor
    Northeastern University
    Email

  • Brian Tracey
    Research Assistant Professor
    Tufts University
    Email

  • Lili He
    Staff
    Massachusetts General Hospital
    Email

  • Alyssa White
    Staff
    Massachusetts General Hospital
    Email

  • Richard Moore
    Staff
    Massachusetts General Hospital
    Email


Students

  • Limor Eger
    Boston University
  • Zachary Sun
    Boston University
  • Oguz Semerici
    Tufts University
  • Pengchong Jin
    Purdue University
  • Eri Haneda
    Purdue University
  • Ke Chen
    Boston University
  • Kirill Trapeznikov
    Boston University
  • Zhengmin Li
    Rensselaer Polytechnic Institute
  • Joe Wang
    Boston University
  • Jing Qian
    Boston University
  • D. Motamed Vaziri
    Boston University
  • Teresa Mao
    Northeastern University
  • Mustafa Ayazoglu
    Northeastern University
  • Binlong Li
    Northeastern University
  • Caglayan Dicle
    Northeastern University
  • Oliver Lehmann
    Northeastern University
  • Ziyan Wu
    Rensselaer Polytechnic Institute
  • Christopher Lo
    Tufts University
  • Rana Hanocka
    Rensselaer Polytechnic Institute
  • Behavior Subtraction P. M. Jodoin, V. Saligrama, J. Konrad IEEE Trans. on Image Processing 2012.
  • Video Anomaly Detection Based on Local Statistical Aggregates V. Saligrama, Z. Chen CVPR 2012.
  • Limited view angle iterative CT reconstruction S. J. Kisner, E. Haneda, C. Bouman, S. Skatter, M. Kourinny and S. Bedford in Proc. SPIE, vol. 8296, 2012. p. 82960F 2012.
  • Graph Constrained Group Testing M. Cheraghchi, A. Karbasi, S. Mohajer, V. Saligrama IEEE Transaction on Information Theory 2012.
  • Multi-Stage Classifier Design K. Trapeznikov, V. Saligrama, D. Castanon ACML 2012.
  • Real-Time Activity Search of Surveillance Video G. Castanon, V. Saligrama, P. M. Jodoin, A. Caron IEEE Advanced Video and Signal-Based Surveillance (AVSS) 2012.
  • Boolean Compressed Sensing and Noisy Group Testing G. Atia, V. Saligrama IEEE Trans. On Information Theory 2012.
  • Dynamic Context for Tracking Behind Occlusions F. Xiong, O. Camps and M. Sznaier Proc. ECCV, to appear 2012.
  • A Convex Optimization Approach to Synthesizing Bounded Complexity l∞ Filters F. Blanchini and M. Sznaier IEEE Trans. Aut. Contr, Vol 57, (1), pp. 216 – 221 2012.
  • Fast Algorithms for Structured Robust Principal Component Analysis Camps, Octavia IEEE CVPE 2012 Proceedings 2012.