Documentation

This is machine translation

Translated by Microsoft
Mouseover text to see original. Click the button below to return to the English verison of the page.

Note: This page has been translated by MathWorks. Please click here
To view all translated materals including this page, select Japan from the country navigator on the bottom of this page.

wden

Automatic 1-D de-noising

Syntax

XD = wden(X,TPTR,SORH,SCAL,N,'wname')
XD = wden(C,L,TPTR,SORH,SCAL,N,'wname')
XD = wden(W,'modwtsqtwolog',SORH,'mln',N,WNAME)
[XD,CXD] = wden(...)
[XD,CXD,LXD] = wden(...)
[XD,CXD,LXD,THR] = wden(...)
[XD,CXD,THR] = wden(...)

Description

wden is a one-dimensional de-noising function.

wden performs an automatic de-noising process of a one-dimensional signal using wavelets.

XD = wden(X,TPTR,SORH,SCAL,N,'wname') returns a de-noised version XD of input signal X obtained by thresholding the wavelet coefficients.

TPTR character vector contains the threshold selection rule:

  • 'rigrsure' uses the principle of Stein's Unbiased Risk.

  • 'heursure' is an heuristic variant of the first option.

  • 'sqtwolog' for the universal threshold 2ln(·)

  • 'minimaxi' for minimax thresholding (see thselect for more information)

SORH ('s' or 'h') is for soft or hard thresholding (see wthresh for more information).

SCAL defines multiplicative threshold rescaling:

'one' for no rescaling

'sln' for rescaling using a single estimation of level noise based on first-level coefficients

'mln' for rescaling done using level-dependent estimation of level noise

Wavelet decomposition is performed at level N and 'wname' is a character vector containing the name of the desired orthogonal wavelet (see wmaxlev and wfilters for more information).

XD = wden(C,L,TPTR,SORH,SCAL,N,'wname') returns the same output arguments, using the same options as above, but obtained directly from the input wavelet decomposition structure [C,L] of the signal to be de-noised, at level N and using 'wname' orthogonal wavelet.

XD = wden(W,'modwtsqtwolog',SORH,'mln',N,WNAME) returns the denoised signal obtained by operating on the MODWT transform matrix W, where W is the output of MODWT. You must use the same wavelet in both modwt and wden.

[XD,CXD] = wden(...) returns the denoised wavelet coefficients. For DWT denoising, CXD is a vector (see wavedec). For MODWT denoising, CXD is a matrix with N+1 rows (see modwt). The number of columns is equal to the length of the input signal X.

[XD,CXD,LXD] = wden(...) returns the number of coefficients by level for DWT denoising. See wavedec for details. The LXD output is not supported for MODWT denoising. The additional output arguments [CXD,LXD] are the wavelet decomposition structure (see wavedec for more information) of the de-noised signal XD.

[XD,CXD,LXD,THR] = wden(...) returns the denoising thresholds by level for DWT denoising.

[XD,CXD,THR] = wden(...) returns the denoising thresholds by level for MODWT denoising when you specify 'modwtsqtwolog'.

Examples

collapse all

In this example, three different denoising techniques are applied to a noisy signal. Plots comparing results are generated. The thresholds produced by each technique are also shown.

First, to ensure reproducibility of results, set a seed that will be used to generate the random noise.

rng('default')

Create a signal consisting of a 2-Hz sine wave with transients at 0.3 and 0.72 seconds. Add randomly generated noise to the signal and plot the result.

N = 1000;
t = linspace(0,1,N);
x = 4*sin(4*pi*t);
x = x - sign(t-0.3) - sign(0.72-t);
sig = x + 0.5*randn(size(t));
plot(t,sig)
title('Signal')

Using the sym8 wavelet, perform a level 5 wavelet decomposition of the signal and denoise it by applying three different threshold selection rules to the wavelet coefficients: SURE; minimax; Donoho and Johnstone's universal threshold with level-dependent estimation of the noise. In each case, apply hard thresholding.

lev = 5;
wname = 'sym8';
[dnsig1,c1,l1,threshold_SURE] = wden(sig,'rigrsure','h','mln',lev,wname);
[dnsig2,c2,l2,threshold_Minimax] = wden(sig,'minimaxi','h','mln',lev,wname);
[dnsig3,c3,l3,threshold_DJ] = wden(sig,'sqtwolog','h','mln',lev,wname);

Plot and compare the three denoised signals.

subplot(3,1,1)
plot(t,dnsig1)
title('Denoised Signal - SURE')
subplot(3,1,2)
plot(t,dnsig2)
title('Denoised Signal - Minimax')
subplot(3,1,3)
plot(t,dnsig3)
title('Denoised Signal - Donoho-Johnstone')

Compare the thresholds applied at each detail level for the three denoising methods.

threshold_SURE
threshold_SURE = 

    0.8079    1.1448    1.3915    0.9816    0.8667

threshold_Minimax
threshold_Minimax = 

    1.0522    1.0881    1.2237    1.3366    1.1985

threshold_DJ
threshold_DJ = 

    1.7644    1.8247    2.0520    2.2413    2.0097

In this example, a signal is denoised using the DWT and MODWT. Plots comparing results are generated. The thresholds used in each case are also shown.

First, to ensure reproducibility of results, set a seed that will be used to generate random noise.

rng('default')

Create a signal consisting of a 2-Hz sine wave with transients at 0.3 and 0.72 seconds. Add randomly generated noise to the signal and plot the result.

N = 1000;
t = linspace(0,1,N);
x = 4*sin(4*pi*t);
x = x - sign(t-0.3) - sign(0.72-t);
sig = x + 0.5*randn(size(t));
plot(t,sig)
title('Signal')

Using the db2 wavelet, perform a level 3 wavelet decomposition of the signal and denoise it using Donoho and Johnstone's universal threshold with level-dependent estimation of the noise. Obtain denoised versions using DWT and MODWT and soft thresholding.

wname = 'db2';
lev = 3;
[xdDWT,c1,l1,threshold_DWT] = wden(sig,'sqtwolog','s','mln',lev,wname);
[xdMODWT,c2,threshold_MODWT] = wden(sig,'modwtsqtwolog','s','mln',lev,wname);

Plot and compare the results.

subplot(2,1,1)
plot(t,xdDWT)
title('DWT Denoising')
subplot(2,1,2)
plot(t,xdMODWT)
title('MODWT Denoising')

Compare the thresholds applied in each case.

threshold_DWT
threshold_DWT = 

    1.7783    1.6876    2.0434

threshold_MODWT
threshold_MODWT = 

    1.2760    0.6405    0.3787

In this example, a blocky signal is denoised using the Haar wavelet with DWT and MODWT denoising. Plots are generated comparing results. Metrics comparing the original with denoised versions are also produced.

First, to ensure reproducibility of results, set a seed that will be used to generate random noise.

rng('default')

Generate a signal and a noisy version with the square root of the signal-to-noise ratio equal to 3. Plot and compare each.

[osig,nsig] = wnoise('blocks',10,3);
plot(nsig,'r')
hold on
plot(osig,'b')
legend('Noisy','Original')

Using the Haar wavelet, perform a level 6 wavelet decomposition of the noisy signal and denoise it using Donoho and Johnstone's universal threshold with level-dependent estimation of the noise. Obtain denoised versions using DWT and MODWT and soft thresholding.

wname = 'haar';
lev = 6 ;
[xdDWT,c1,l1] = wden(nsig,'sqtwolog','s','mln',lev,wname);
[xdMODWT,c2] = wden(nsig,'modwtsqtwolog','s','mln',lev,wname);

Plot and compare the original, noise-free version of the signal with the two denoised versions.

figure
plot(osig,'b')
hold on
plot(xdDWT,'r--')
plot(xdMODWT,'k-.')
legend('Original','DWT','MODWT')
hold off

Calculate the L2 and L-infinity norms of the difference between the original signal and the two denoised versions.

L2norm_original_DWT = norm(abs(osig-xdDWT),2)
L2norm_original_DWT = 36.1194
L2norm_original_MODWT = norm(abs(osig-xdMODWT),2)
L2norm_original_MODWT = 14.5987
LInfinity_original_DWT = norm(abs(osig-xdDWT),Inf)
LInfinity_original_DWT = 4.7181
LInfinity_original_MODWT = norm(abs(osig-xdMODWT),Inf)
LInfinity_original_MODWT = 2.9655

Algorithms

The underlying model for the noisy signal is basically of the following form:

s(n)=f(n)+σe(n)

where time n is equally spaced.

In the simplest model, suppose that e(n) is a Gaussian white noise N(0,1) and the noise level σ a is supposed to be equal to 1.

The de-noising objective is to suppress the noise part of the signal s and to recover f.

The de-noising procedure proceeds in three steps:

  1. Decomposition. Choose a wavelet, and choose a level N. Compute the wavelet decomposition of the signal s at level N.

  2. Detail coefficients thresholding. For each level from 1 to N, select a threshold and apply soft thresholding to the detail coefficients.

  3. Reconstruction. Compute wavelet reconstruction based on the original approximation coefficients of level N and the modified detail coefficients of levels from 1 to N.

More details about threshold selection rules are in Wavelet Denoising and Nonparametric Function Estimation, in the User's Guide, and in the help of the thselect function. Let us point out that

  • The detail coefficients vector is the superposition of the coefficients of f and the coefficients of e, and that the decomposition of e leads to detail coefficients that are standard Gaussian white noises.

  • Minimax and SURE threshold selection rules are more conservative and are more convenient when small details of function f lie in the noise range. The two other rules remove the noise more efficiently. The option 'heursure' is a compromise.

In practice, the basic model cannot be used directly. This section examines the options available, to deal with model deviations. The remaining parameter scal has to be specified. It corresponds to threshold rescaling methods.

  • Option scal = 'one' corresponds to the basic model.

  • In general, you can ignore the noise level that must be estimated. The detail coefficients CD1 (the finest scale) are essentially noise coefficients with standard deviation equal to σ. The median absolute deviation of the coefficients is a robust estimate of σ. The use of a robust estimate is crucial because if level 1 coefficients contain f details, these details are concentrated in few coefficients to avoid signal end effects, which are pure artifacts due to computations on the edges.

  • The option scal = 'sln' handles threshold rescaling using a single estimation of level noise based on the first-level coefficients.

  • When you suspect a nonwhite noise e, thresholds must be rescaled by a level-dependent estimation of the level noise. The same kind of strategy is used by estimating σlev level by level. This estimation is implemented in the file wnoisest, which handles the wavelet decomposition structure of the original signal s directly.

  • The option scal = 'mln' handles threshold rescaling using a level-dependent estimation of the level noise.

References

Antoniadis, A.; G. Oppenheim, Eds. (1995), Wavelets and statistics, 103, Lecture Notes in Statistics, Springer Verlag.

Donoho, D.L. (1993), “Progress in wavelet analysis and WVD: a ten minute tour,” in Progress in wavelet analysis and applications, Y. Meyer, S. Roques, pp. 109–128. Frontières Ed.

Donoho, D.L.; I.M. Johnstone (1994), “Ideal spatial adaptation by wavelet shrinkage,” Biometrika, Vol. 81, pp. 425–455.

Donoho, D.L. (1995), “De-noising by soft-thresholding,” IEEE Trans. on Inf. Theory, 42 3, pp. 613– 627.

Donoho, D.L.; I.M. Johnstone, G. Kerkyacharian, D. Picard (1995), “Wavelet shrinkage: asymptotia,” Jour. Roy. Stat. Soc., series B, Vol. 57, No. 2, pp. 301–369.

Extended Capabilities

Introduced before R2006a

Was this topic helpful?