Purpose

Improvised Explosive Devices (IEDs) are one
of today’s major threats that anti-terrorism actors
have to cope with. We propose image processing
detection solutions applied to a vehicle-mounted
camera which make it possible to warn the
driver of the presence of changes, i.e. potential
IEDs, on his itinerary. The detection method is
based on GPS-referenced pictures which are
registered and normalized in order to be compared for change detection. Possible applications range from the security of troops
deployed overseas to the safety of railways
or airport runways.

Skills

• Image registration
• High-performance video stabilization
• Image change detection
• Camera calibration
• GPU-based real-time image processing
• FPGA-based embedded image processing.

Facilities

• ISL private test facilities
• Fully equipped test vehicle (digital cameras with motorized zoom mounted on motorized head,
• DGPS-RTK positioning system with
  1-cm accuracy at 20 Hz, IMU and electronic compass, telemetry, rugged military laptop, multi-GPU embedded workstation).

Results

We have shown that potential IED detection along frequently used itineraries can be carried out by detecting changes in the pictures taken in the same location but at different times. In practice, the pictures to be compared are not taken in the very same location, have different frame alignments and different exposures. Even in difficult varying illumination conditions (different position of the sun, shadows, rain), we achieve by now encouraging detection results like:

• detection of objects as small as a soda can at about 200 m
• detection of otherwise unnoticeable vehicle tracks on grass
• detection of scraped ground zones corresponding to buried mines.