wavegrow: Interactive graphical tool to grow a wavelet synthesis

Description

Use mouse to select which wavelets to enter a wavelet synthesis, continually plot the reconstruction and the wavelet tableaux.

Usage

wavegrow(n = 64, filter.number = 10, family = "DaubLeAsymm", type = "wavelet",
    random = TRUE, read.value = TRUE, restart = FALSE)

Value

The final tableaux.

Arguments

n: Number of points in the decomposition
filter.number: The wavelet filter.number to use, see filter.select
family: The wavelet family to use in the reconstruction
type: If "wavelet" then carry out the regular wavelet transform, otherwise if "station" do the nondecimated transform.
random: If TRUE then iid Gaussian coefficients are inserted into the tableaux. If FALSE and read.value=TRUE then the user is promoted for a value, otherwise the value 1 is inserted into the tableaux at the selected point.
read.value: If TRUE then a value is read and used to insert that size of wavelet coefficient at the selected point. If FALSE then a coefficient of size 1 is inserted.
restart: If TRUE then after a coefficient has been inserted, and plots done, the next selection causes all the coefficients to be reset to zero and a single coefficient inserted. This actually has the overall action of being able to select a coefficient location and view the size and shape of the wavelet produced.

Author

G P Nason

Details

This function can perform many slightly different actions. However, the basic idea is for a tableaux of wavelet coefficients to be displayed in one graphics window, and the reconstruction of those coefficients to be displayed in another graphics window.

Hence, two graphics windows, capable of plotting and mouse interaction (e.g. X11, windows or quartz) with the locator function, are required to be active.

When the function starts up an initial random tableaux is displayed and its reconstruction.

The next step is for the user to select coefficients on the tableaux. What happens next specifically depends on the arguments above. By default selecting a coefficient causes that coefficient scale and location to be identified, then a random sample is taken from a N(0,1) random variable and assigned to that coefficient. Hence, the tableaux is updated, the reconstruction with the new coefficient computed and both are plotted.

If type="wavelet" is used then decimated wavelets are used, if type="station" then the time-ordered non-decimated wavelets are used.

If random=FALSE then new values for the coefficients are either selected (by asking the user for input) if read.value=TRUE or the value of 1 is input.

If restart=TRUE then the function merely displays the wavelet associated with the selected coefficient. Hence, this option is useful to demonstrate to people how wavelets from different points of the tableaux have different sizes, scales and locations.