ENLARGE Choose Target Entropy (bits) 4 3 2 1.5 1 .75 .5 .25 |
Using a compression algorithm
written by Chris Brislawn of Los Alamos' CIC-3, I performed eight compression
experiments on each of the thirteen signals.
The spectrogram on the right has been compressed with a target entropy of 1 (measured in bits). You can explore what the other levels of compression have done to the spectrogram by choosing the appropriate target entropy to the left. Notice how the lightning signature has not been affected. Instead, the compression has effectively removed many of the time/frequency bins where background noise occurs (the black vertical strips). This is acheived through time/frequency-localization. This characteristic of wavelet compression has preserved the parameters from each lightning signature (contrast, TEC, lag, etc.) which we would like to archive. We can see how well they were preserved by examining the relative error. View
the effect of compression on diffuse signals.
|
