Faculty of Engineering and Science
The International Doctoral School of Technology and Science

PhD Course

Time-Frequency Analysis


During the November session I showed a number of examples using the Uvi_Wave toolbox. Links to the m-files with the examples are below, with brief descriptions.


As part of the course you are asked to do some exercises. The first two exercises are fairly well defined, the third one is open-ended. The fourth one requires some care in interpretation, and shows how one can manipulate the plots.
  1. In each of the four examples above experiment with the various parameters. Change the length of the filter. Note that for the Daubechies filters the length must be even. Try to change to other filters. Possibilities include symlets, see
    help symlets
    and the CDF-family described in the lectures, see
    help wspline
    Try changing the level of the best level basis used for the first plot in the first three examples (see comments in the m-files). In the wavelet packet with cost function lp-norm try changing the value of the p-parameter. In the original m-file it is 1.5.
  2. Replace the signal in the examples with another synthetic signal.
  3. Use the experience you have gained in the two exercises above to analyse some real-life signals.
  4. The file tfp4.m uses a transformation of the wavelet packet coefficients, before the plot is made. They are in line 58
    Y2 = log(abs(Y2)+1);
    and in line 72,
    Try replacing the power 2 by 1. Try other transformations. Compare the plots in the different cases, when you repeat the above computatins in the examples with this modified plotter.

This page is maintained by Arne Jensen.
Updated November 18, 2003, Arne Jensen.