Human Identification at a Distance (HID)
Human identification from facial features has been studied primarily
using imagery from visible video cameras. Thermal imaging sensors are
one of the most innovative emerging techonologies in the market.
Fueled by ever lowering costs and improved sensitivity and resolution,
our sensors provide exciting new oportunities for biometric
identification. As part of our involvement in this effort, Equinox is
collecting an extensive database of face imagery in the following
broadband-visible/LWIR (8-12 microns), MWIR (3-5 microns), SWIR
(0.9-1.7 microns). This data collection is
made available on this Web Site for experimentation
and statistical performance evaluations. During Phase I, initial
performance evaluations will be made to assess the utility of thermal
infrared images for face recognition, detection and tracking.
The gray value response of a thermal sensor depends on the temperature and
emisivity of the target, as well as intrinsic properties of the focal plane
array. In order to translate such responses into invariant physical flux
measurements, it is necessary to calibrate the sensor with respect to a black
body source. This allows us to compare data aquired under various conditions
with respect to a single standard. With this in mind, it is important to
establish formal protocols for collection of thermal infrared imagery.
A key problem for face recognition from visible imagery has been accurate
identification under variable illumination conditions. The visible light
impigent on a camera sensor is a product of the reflectivity of the subject's
skin and the light incident on thereon. Therefore, it is impossible to
separate intrinsic properties of the subject from those of the ambiant
illumination conditions. On the other hand, thermal emission from skin is an
intrinsic property, independent of illumination. In order to better exploit
this invariance for face recognition, Equinox is conducting thorough data
collections and evaluation efforts. Validation trials are being conducted
to compare recognition performance of visible and thermal imagery.
In support of these trials, a coregistered visible/LWIR camera sensor is
being used to collect image pairs under identical conditions.
Additionally, boresighted SWIR and MWIR imagery is being aquired.
This allows us to test algorithm performance under controlled conditions.