Improving the Efficiency of Joint SWT/WPT Analysis by Using Different Wavelet Mothers

Previous studies related to joint analysis by using Discrete Wavelet Transform and Wavelet Packet transform relying on Daubechies filters of length 28 were considered as a starting point for this paper. Sensibilities to harmonic pollution of levels from Stationary Wavelet Trees relying on Daubechies filters of 28 components, of 40 components and respectively of a truncated version of the last ones (containing only 30 components) were determined and compared. Their performances were evaluated in order to be used in another type of joint analysis, involving Wavelet Packet trees. The same filters were examined afterward relative to the most significant harmonic energies weights in the terminal nodes of a wavelet packet tree with 7 levels used to decompose signals with 512 components per period. Runtimes were evaluated for all types of wavelet transforms (Discrete, Stationary and Packet) and recommendations relative to "best usability" scenarios were issued based on the performed analysis.