Abstract
The ability to extract and describe the salient features represents the most difficult task of
the aircraft recognition systems. In this paper, an aircraft recognition system based on the
isodensity lines associated with the three-dimensional reflectivity of the model is illustrated.
A recognition algorithm is described, which uses combination of 2-D moments in threedimensional
isodensity maps to represent the aircraft as a feature vector containing 2L elements.
Using a distance-weighted k-nearest neighbour rule as a classifier, the algorithm achieves
a highly recognition rate when applied to 336 test images that represent six aircraft models. The
same procedures are applied using the first two components of invariant moments, and the first
system was found superior to this system with about 15% using the same number of isodensity
lines.