| Author: | Filip Wasilewski |
|---|---|
| Contact: | filipwasilewski@gmail.com |
| Version: | 0.1.2 |
| Status: | alpha |
| Date: | 2006-03-25 17:30 |
| License: | MIT |
| Changelog: | CHANGES.txt |
Abstract
PyWavelets is a Python module for calculating Simple and Inverse Discrete Wavelet Transform, as well as Wavelet Packets and Stationary Wavelet Transform. This document is User Guide to PyWavelets.
PyWavelets was developed and tested under Python 2.4. Most of the code is written in C with Python bindings written in Pyrex.
PyWavelets module works well with any recent numeric module implementing array interface, like numpy (recommended), Numeric or numarray. It can also be used standalone utilising standard array.array builtin Python module.
Current source release is available for download from:
PyWavelets-0.1.2.zip
The precompiled binary version for Windows can be downloaded from:
PyWavelets-0.1.2.win32-py2.4.exe
PyWavelets was developed with MinGW C compiler, Pyrex and Python 2.4 under WindowsXP.
If You are using Python 2.4 on Windows just download and execute the binary version for Windows.
If You are installing from source try executing setup script:
setup.py install
If You have numpy and matplotlib installed, try running some examples from demo directory to verify installation.
PyWavelets is available under MIT license.
Feel free to contact me directly filipwasilewski@gmail.com. Comments and bug reports (and fixes;) are welcome.
The families() function returns names of available wavelet families.
Currently six wavelet families are implemented with over seventhy wavelets available:
Example:
>>> import pywt >>> print pywt.families() ['haar', 'db', 'sym', 'coif', 'bior', 'rbio', 'dmey']
The wavelist(short_name=None) function returns list of available wavelet names.
If short_name is None, then names of all implemented wavelets is returned, otherwise the function returns names of wavelets from given family name.
Example:
>>> import pywt >>> print pywt.wavelist('coif') ['coif1', 'coif2', 'coif3', 'coif4', 'coif5']
Wavelet(name, filterBank=None) class describes properties of wavelet identified by name. The parameter name must be valid wavelet name from wavelist() list. Otherwise the filterBank must be provided.
Example:
>>> wavelet = pywt.Wavelet('db1') >>> print wavelet Wavelet db1 Family name: Daubechies Short name: db Filters length: 2 Orthogonal: True Biorthogonal: True Orthonormal: False Symmetry: asymmetric >>> print wavelet.dec_lo, wavelet.dec_hi [ 0.70710678 0.70710678] [-0.70710678 0.70710678] >>> print wavelet.rec_lo, wavelet.rec_hi [ 0.70710678 0.70710678] [ 0.70710678 -0.70710678]
The wavefun(level) function can be used to calculates aproximations wavelet function (psi) and associated of scaling function (phi) at given level of refinement.
For an orthogonal wavelet returns scaling and wavelet function.
>>> wavelet = Wavelet('db2') >>> phi, psi = wavelet.wavefun(level=5)
For an biorthogonal wavelet returns scaling and wavelet function both for decomposition and reconstruction.
>>> wavelet = Wavelet('bior1.1') >>> phi_d, psi_d, phi_r, psi_r = wavelet.wavefun(level=5)
See also plots of Daubechies and Symlets wavelet familes generated with wavefun function:
Wavelet transform has recently became very popular when it comes to data analysis, denoising or compression, either 1 or 2 dimensional signals.
The dwt function is used to perform single level, one dimensional Discrete Wavelet Transform.
(cA, cD) = dwt(data, wavelet, mode=MODES.sym)
Length of returned approximation (cA) and details(cD) coefficient arrays depends on selected mode - see dwt_coeff_len:
for all modes except periodization:
len(cA) == len(cD) == floor((len(data) + wavelet.dec_len - 1) / 2)
for periodization mode (MODES.per):
len(cA) == len(cD) == ceil(len(data) / 2)
Example:
>>> import pywt >>> cA,cD = pywt.dwt([1,2,3,4,5,6], 'db1') >>> print cA [ 2.12132034 4.94974747 7.77817459] >>> print cD [-0.70710678 -0.70710678 -0.70710678]
Performs multilevel Discrete Wavelet Transform decomposition of given signal and returns coefficient list in form [cAn cDn cDn-1 ... cD2 cD1].
wavedec(data, wavelet, level=1, mode=MODES.sym)
Example:
>>> import pywt >>> coeffs = pywt.wavedec([1,2,3,4,5,6,7,8], 'db1', level=2) >>> a2, d2, d1 = coeffs >>> print d1 [-0.70710678 -0.70710678 -0.70710678 -0.70710678] >>> print d2 [-2. -2.] >>> print a2 [ 5. 13.]
The dwt_max_level function can be used to compute the maximum usefull level of decomposition for given input data length and wavelet filter length.
dwt_max_level(data_len, filter_len)
The returned value equals:
floor(log(data_len/(filter_len-1))/log(2))
Although the maximum decomposition level can be quite big for long signals, usually smaller values are chosen (5 for 1-dimensional signals).
Example:
>>> import pywt >>> w = pywt.Wavelet('sym5') >>> print pywt.dwt_max_level(input_len = 1000, filter_len = w.dec_len) 6
Depending on the selected signal extension mode, the dwt_coeff_len function calculates the length of dwt coefficients.
dwt_coeff_len(data_len, filter_len, object mode)
For periodization mode this equals:
ceil(data_len / 2)
which is the lowest possible length guaranteeing perfect reconstruction.
For other modes:
floor((data_len + filter_len - 1) / 2)
To handle problem of border distortion while performing DWT, one of several signal extension modes can be selected.
zpd - zero-padpadding - signal is extended by adding zero samples:
0 0 | x1 x2 ... xn | 0 0
cpd - constant-padding - edge values are used:
x1 x1 | x1 x2 ... xn | xn xn
sym - symmetric-padding - signal is extended by mirroring samples:
x2 x1 | x1 x2 ... xn | xn xn-1
ppd - periodic-padding - signal is treated as periodic:
xn-1 xn | x1 x2 ... xn | x1 x2
sp1 - smooth-padding - signal is extended accordin to first derivatives calculated on the edges
DWT performed for these extension modes is slightly redundant, but ensure a perfect reconstruction. To receive the smallest number of coefficients DWT can be computed with periodization mode
Example:
>>> import pywt >>> print pywt.MODES.modes ['zpd', 'cpd', 'sym', 'ppd', 'sp1', 'per']
Notice that you can use either of the following forms:
>>> (a,d) = pywt.dwt([1,2,3,4,5,6], 'db2', 'sp1') >>> (a,d) = pywt.dwt([1,2,3,4,5,6], pywt.Wavelet('db2'), pywt.MODES.sp1)
It is important to note that signal in memory isn't really extended, but the extra values are computed without data copying. This feature can sometimes prevent extra page swap when when handling big data arrays on computers with low physical memory available.
The idwt function reconstruct data from given coefficients by performing Single level Inverse Discrete Wavelet Transform.
idwt(cA, cD, wavelet, mode=MODES.sym, correct_size=0)
Example:
>>> import pywt >>> (cA,cD) = pywt.dwt([1,2,3,4,5,6], 'db2', 'sp1') >>> print pywt.idwt(cA, cD, 'db2', 'sp1') [ 1. 2. 3. 4. 5. 6.]
Notice that one of cA and cD arguments can be None. In that situation the reconstruction of the not None argument will be performed.
Example:
>>> import pywt, numpy >>> (cA,cD) = pywt.dwt([1,2,3,4,5,6], 'db2', 'sp1') >>> A = pywt.idwt(cA, None, 'db2', 'sp1') >>> D = pywt.idwt(None, cD, 'db2', 'sp1') >>> print A + D [ 1. 2. 3. 4. 5. 6.]
Performs multilevel reconstruction of signal from given coefficient list.
waverec(coeffs_list, wavelet, mode=MODES.sym)
coefficient list must be in form like that from wavedec decomposition:
[cAn cDn cDn-1 ... cD2 cD1]
Example:
>>> import pywt >>> coeffs = pywt.wavedec([1,2,3,4,5,6,7,8], 'db2', level=2) >>> print pywt.waverec(coeffs, 'db2') [ 1. 2. 3. 4. 5. 6. 7. 8.]
Direct reconstruction from cefficients.
upcoef(part, coef, wavelet, level=1, take=0)
defines coefficients type:
Example:
>>> import pywt, numpy >>> data = [1,2,3,4,5,6] >>> (cA,cD) = pywt.dwt(data, 'db2', 'sp1') >>> print pywt.upcoef('a', cA, 'db2') + pywt.upcoef('d', cD, 'db2') >>> print pywt.upcoef('a', cA, 'db2') + pywt.upcoef('d', cD, 'db2') [-0.25 -0.4330127 1. 2. 3. 4. 5. 6. 1.78589838 -1.03108891] >>> n = len(data) >>> print pywt.upcoef('a',cA,'db2',take=n) + pywt.upcoef('d',cD,'db2',take=n) [ 1. 2. 3. 4. 5. 6.]
Tree structure simplyfying operations on Wavelet Packet decomposition coefficients. It consists of Node elements.
WaveletPacket(data, wavelet, mode=pywt.MODES.sp1, maxlevel=None)
wp = WaveletPacket(range(16), 'db1', maxlevel=3)
Find node of given path in tree.
If node does not exist yet, it will be created by decomposition of its parent node.
Calls get_node(path) and returns data associated with node under given path.
Calls get_node(path) and sets data of node under given path.
Marks node under given path in tree as ZeroTree root.
If node does not exist yet, it will be created by decomposition of its parent node.
Returns data reconstruction using coefficients from subnodes.
If update is True, then node's data values will be replaced by reconstruction values (also in subnodes).
Returns all nodes from specified level.
Returns non-zero terminal nodes.
Walks tree and calls func on every node - func(node, *args). If func returns True, descending to subnodes will proceed.
Walks tree and calls func on every node starting from bottom most nodes.
WaveletPacket tree node.
Subnodes are called 'a' and 'd', like approximation and detail coefficients in Discrete Wavelet Transform
Path under node is accessible in Wavelet Packet tree.
Data associated with node.
Mark node as root of ZeroTree, which means that current node and all subnodes don't take part in reconstruction (all coefficients equals 0).
Field - like markZeroTree.
Returns chosen subnode.
Performs multilevel Stationary Wavelet Transform.
swt(data, wavelet, level)
Returned list of coefficient pairs is in form [(cA1, cD1), (cA2, cD2), ..., (cAn, cDn)], where n = level
Returns maximum level of Stationary Wavelet Transform for data of given length.
swt_max_level(input_len)
