Kevin Rudd - kerudd@wm.edu

The Wavelet Thumbprint Toolbox - WTPtool

The wavelet fingerprint technique is an algorithm for converting complicated one-dimensional signals into 2D fingerprint-like images.  Features in a one-dimensional signal have a corresponding wavelet fingerprint that is easy to identify visually or can be detected automatically with image processing algorithms.  The wavelet fingerprint technique has been successfully applied to a variety of different applications including feature extraction algorithms that interpret echoes from an ultrasonographic periodontal probe, extract multi-mode waveform properties from a guided wave tomography system, identify deeply-buried delaminations in microchip packaging via acoustic microscopy, and detect chafing of insulation on coaxial cable and twisted pair wiring.  In each case, the wavelet fingerprint (WFP) technique was successful where traditional time domain signal processing had failed.

We have created an interactive MATLAB based wavelet thumbprint toolbox to analyze one-dimensional signals.  With this toolbox, one can explore different wavelet fingerprint parameters such as the mother wavelet and the number and width of the fingerprint ridge lines.  It can be used to visually explore wavelet fingerprints and how they correspond to features in the original 1D signal; some of which may not be visible to the human eye.

Sample Wavelet Thumbprints

WTPtool