The objective of this thesis is to estimate stable harmonic signatures for acoustic
sources. A MATLAB program called ACIDSTOOL is developed in this thesis to process
measured data from acoustic sources in the ACIDS (Acoustic-seismic Classification
Identification Data Set) database. The harmonic signatures consist of fundamental
frequency and harmonic amplitudes. The harmonics are caused by rotating machinery of
the engines of vehicles. Reliable harmonic signatures are important for building
classifiers that distinguish the type of moving vehicles.
We develop two methods based on Harmonic Line Association (HLA) to estimate
the harmonic signatures. The ACIDSTOOL program has a Graphical User Interface (GUI)
that enables exploration and visualization of the large amount of data in the ACIDS
database. We include the results from processing the ACIDS database with the
ACIDSTOOL program.