Potential asymmetric threats at short range in complex environments need to be identified quickly during coastal operations.
Laser range profiling is a technology that has the potential to shorten the OODA loop (Orient, Observe, Detect,
Act) by performing automatic characterisation of targets at large distance. The advantages of non-cooperative target
recognition with range profiles are: (a) a relatively short time on target is required, (b) the detection range is longer than
in the case of passive observation technologies such as IRST, and (c) characterisation of range profiles is possible at any
aspect angle. However, the shape of a range profile depends strongly on aspect angle. This means that a large data set is
necessary of all expected targets with reference profiles on a very dense aspect angle grid. Analysis of laser range profiles
can be done by comparing the measured profile with a database of laser range profiles obtained from 3D models of
possible targets. An alternative is the use of a profile database from one or several measurement campaigns. A prerequisite
for this is the availability of enough measured profiles of the appropriate targets, for many aspect angles. Comparison
of measured laser range profiles with a reference database can be performed using, e.g., formal statistical correlation
techniques or histogram dissimilarity techniques.
In this work, a field trial has been conducted to validate the concept of identification by using a laser range profiling
system with a high bandwidth receiver and short laser pulses. The field trial aimed at characterization of sea-surface
targets in a coastal/harbour environment. The targets ranged from pleasure boats like sailing boats, jet skis, and speed
boats to professional vessels like barges, cabin boats, and military vessels, all ranging from 3 to 30 meters in length. We
focus on (a) the use of a reference database generated via 3D target models, and (b) the use of a reference database of
measured laser range profiles. A variety of histogram dissimilarity measures was examined in order to enable fast and
reliable classification algorithms.