Warning

1
Warning Indicator Systems WISE lab

• Lab Director Professor Richard Mammone
• Faculty Christine Podilchuck, Joseph Wilder
• Students Anand Doshi, Aparna Krishnamoorthy,
• Robert Utama
• http//www.caip.rutgers.edu/wiselab

2

• Enhancement of Signals
• filtering
• deconvolution
• noise reduction
• segmentation

• Identification of Materials/People
• radioactive material
• plastic explosives
• person identity

• Positional Awareness
• sensor location
• target location
• target tracking
• target prediction

3

• Projects
• Radioactive Isotope Identification
• Positional Awareness
• Face Recognition
• Speaker Recognition
• Explosives Detection

4
The Rutgers CAIP Vigilant Decision Machine (VDM)
detects and Identifies materials used in
radiological weapons.

• The VDM analyses data captured from moving
  vehicles by the Princeton Plasma Physics
  Laboratorys Miniature Integrated Nuclear
  Detection System (MINDS).
• The MINDS/VDM system can detect small quantities
  of materials hidden in vehicles,even when
  shielded by lead, and can distinguish between
  dangerous and benign materials.
• A proof-of-principal system is operational and
  has been tested on two of the five materials
  (Cs-137 and Co-60) listed in draft ANSI Standard
  N42.35 2003, and a third material Am-241, which
  was not included in the standard, because it was
  deemed too difficult to detect.
• Our system shows promise of exceeding all of
  the requirements in ANSI N42.35 2003 and adding
  additional desirable capabilities.

5
Data Acquisition
Initial Data Collection Experiment

6
Miniature Integrated Nuclear Detection System
Feature Extraction
Classification
Sensors
Preprocessing
WISE LAB
PPPL
7
PreprocessingBackground and Cesium 137
8
The need for Preprocessing
9
Feature Extraction
10
Neural Tree Network (NTN)
11
Difficult Sample of Co-60 Correctly Identified by
our System
12
Target Tracking Algorithm for Air Deliverable
Acoustic Sensor (ADAS) The objective of the
project is to devise a target tracking algorithm
for the ADAS (pictured below) to locate a target
in a dynamic, noisy and mobile environment
By using several ADAS devices, and an appropriate
algorithm, it is possible to triangulate a
position of a target by analyzing the sound it
emits.
13
Geometry used in the Government Test Three nodes
(each node is an ADAS) spread in three different
locations. A target vehicle will move down the
road equipped with the GPS. The sensors will pick
up the acoustic data and supply it to the WISE
lab algorithm which will map the vehicle's path.
14
Other Projects Speaker Recognition, Explosive
Detection, Face Recognition
Feature Extraction
Classification
Sensors
Preprocessing
WISE LAB
15
Summary and Conclusions

• Rutgers Wise Lab has developed real world
  biometric systems.
• Wise Lab has developed real world material
  identification systems.
• System Engineering including integration of
  hardware and software has been performed by
  WISE Lab.
• The Neural Tree Network has been shown to
  correctly identify people and material when
  little data is available and under noisy
  conditions.
• Appropriate preprocessing and feature selection
  are important design considerations used by the
  WISE Lab.