Abstract
A composite signal is formed by superposition of several wavelets. The shapes of the constituent wavelets need not in general be identical. These composite signals occur in diverse areas such as speech, radar, sonar and electrophysiology. The decomposition of a composite signal aims at obtaining the arrival times or the epochs of the component wavelets. It is well- known that the Fourier transform of a wavelet is related to the epoch. This property is made use of to accomplish composite signal decomposition. In this technique, an appropriate window function is chosen in the high frequency region to perform a band-pass operation on the signal. The technique does not require a priori knowledge of the constituent wavelets for the extraction of the epochs. However, when the wavelets have identical shape then their relative amplitudes and waveshapes can also be obtained. Several examples are given to illustrate the principle of this new decomposition technique.
Indexing Terms: