CThresh - shrinkage of real signals with complex-valued wavelets


CThresh is an add-on for WaveThresh that allows the use of complex-valued Daubechies' wavelets for estimating real signals. Both non-Bayesian (multiwavelet style) and empirical Bayesian shrinkage rules are used.

A paper describing the methods used in Barber & Nason. A preprint version of this paper is available.

The basic WaveThresh3 distribution is a prerequisite for CThresh. You must have WaveThresh3 loaded to be able to run CThresh. WaveThresh3 itself requires R - see the wavethresh web pages for more details.

The following software and help pages were written by Stuart Barber.

The CThresh software can be downloaded here:

Once you have downloaded the distribution file, you can extract the contents with the commands (on a UNIX / Linux system)
gunzip cthresh.tar.gz
tar -xvf cthresh.tar
On Windows, an archiving utility like WinZip or EasyZip should be able to extract the necessary files.

Installation is described in the README file in the distribution (note that Windows users will need to be able to compile the C code, sorry). For the empirical Bayes method to work at its fasted, you will need access to the FORTRAN NAG library.

Cthresh has a number of help pages. They are included in the distribution file above, or you can look at them separately:


[University of Leeds] [School of Mathematics] [Department of Statistics] [Home page] [Research page]
Last modified: Wed Nov 3 08:20:09 GMT 2004