pieflab
所属分类：matlab编程
开发工具：matlab
文件大小：111KB
下载次数：68
上传日期：2007-12-02 12:02:38
上 传 者：
weiw33
说明： ％ PIEFLAB Main Directory
％ ----------------------
％
％ .m - files
％ ----------
％ Contents.m : this file
％ startup.m : startup file: sets Matlab path executed automatically when
％ Matlab command is performed in this directory
％
％ subdirectories
％ --------------
％ General/ : general matlab commands
％ MRF/ : Markov Random Field: Bayesian algorithm for images
％ Noise/ : noise generation, density/distribution functions
％ Tests/
％ Threshold/ : threshold procedures (includes threshold assessment)
％ WT/ : Wavelet Transform routines
％ Poisson/ : Poisson intensity estimation routines
％ Copyright (c) Maarten Jansen
％ K.U.Leuven - Flanders (Belgium)
％
％ This software is part of PiefLab and is copyrighted material. More info on
％ copyright policy is available on:
％ www.cs.kuleuven.ac.be/~maarten/software/
( PIEFLAB Main Directory----------------------. M- files---------- Contents.m: this file startup.m: startup file: sets Matlab path executed automatically when Matlab command is performed in this directory subdirectories-------------- General /: general matlab commands MRF /: Markov Random Field: Bayesian algorithm for images Noise /: noise generation, density/distribution functions Tests/Threshold /: threshold procedures (includes threshold assessment) WT /: Wavelet Transform routines Poisson /: Poisson intensity estimation routines Copyright (c) Maarten Jansen KULeuven- Flanders (Belgium) This software is part of PiefLab and is copyrighted material. More info on copyright policy is available on: www.cs.kuleuven.ac.be/ ~ maarten/software /)
文件列表:
PiefLab (0, 2005-10-05)
PiefLab\Contents.m (844, 2004-01-14)
PiefLab\copyright (1221, 2001-06-16)
PiefLab\General (0, 2004-11-01)
PiefLab\General\column.m (431, 2003-09-08)
PiefLab\General\Contents.m (671, 2003-09-08)
PiefLab\General\div.m (319, 2001-08-29)
PiefLab\General\ffff (0, 2003-09-08)
PiefLab\General\gramschmidt.m (519, 2000-11-19)
PiefLab\General\grey.m (207, 2003-03-20)
PiefLab\General\row.m (421, 2000-11-19)
PiefLab\General\shift.m (512, 2004-11-01)
PiefLab\linstartup.m (684, 2004-01-13)
PiefLab\MRF (0, 2002-04-17)
PiefLab\MRF\condprob.m (512, 2000-11-21)
PiefLab\MRF\condprob1.m (530, 2000-11-19)
PiefLab\MRF\condprob2.m (540, 2000-11-19)
PiefLab\MRF\Contents.m (880, 2000-11-22)
PiefLab\MRF\isingpot.m (838, 2000-11-21)
PiefLab\MRF\isingpotupdate.m (1049, 2000-11-21)
PiefLab\MRF\metropolisIsing.m (899, 2002-04-17)
PiefLab\MRF\MPLEising.m (952, 2000-11-21)
PiefLab\MRF\PLising.m (552, 2001-06-15)
PiefLab\Noise (0, 2005-10-18)
PiefLab\Noise\Contents.m (1275, 2005-03-31)
PiefLab\Noise\cumchisquare.m (722, 2004-05-24)
PiefLab\Noise\cumgauss.m (400, 2000-11-19)
PiefLab\Noise\exptimes.m (727, 2004-02-02)
PiefLab\Noise\gauss.m (404, 2003-08-25)
PiefLab\Noise\invcumgauss.m (504, 2004-11-02)
PiefLab\Noise\MAD1.m (463, 2005-10-18)
PiefLab\Noise\makenoisy.m (1172, 2005-10-18)
PiefLab\Noise\mean2.m (293, 2001-11-19)
PiefLab\Noise\median1.m (379, 2000-11-19)
PiefLab\Noise\medianfilt.m (491, 2004-02-17)
PiefLab\Noise\psnr.m (409, 2005-09-08)
PiefLab\Noise\psnr2.m (413, 2005-09-08)
PiefLab\Noise\randcauchy.m (1128, 2004-06-02)
PiefLab\Noise\randchi2.m (894, 2004-11-02)
PiefLab\Noise\randexp.m (1012, 2004-11-02)
... ...
Note on filters in wavelet transform
------------------------------------
a filter is characterized by two variables:
1) a vector with filter coefficients
2) an integer indicating the shift of the first entry in the vector.
e.g. if h = [1 2 3 4]
and sh = -2;
then the first entry (h(1) = 1) should be shifted two positions to the left
upon using this vector in a filter-operation.
How to determine the exact shift variables?
-------------------------------------------
For perfect reconstruction, it is of course important to have the right values
for the shift variables.
If ht and gt are the dual filters (for analysis) and h and g are the primal
filters (for synthesis or reconstruction), then one of the shifts is free.
So, we put for instance sh = 0.
Perfect reconstruction requires that h and ht have double shift orthogonality.
We should shift ht such that it matches this condition w.r.t. h.
From the (sufficient) condition for perfect reconstruction:
gt_i = (-1)^i * h_{1-i},
which is in matlab (due to the fact that indexing always starts at 1):
gt(i) = gt_{i-1} = (-1)^(i-1) * h_{2-i} = (-1)^(i-1) * h(3-i),
we see that
sgt = first[gt] - 1 = 3 - last[h] - 1 = 2 - (first[h] + length(h) - 1)
= 2 - (sh + length(h))
sgt = 2 - sh - length(h);
sg = 2 - sht - length(ht);
(example: haar filters have length 2: they enter the wavelet transform with all
shifts equal to zero)
The paramter alfa in the reconstruction from the non-decimated transform
------------------------------------------------------------------------
(IRT_BIO.m)
近期下载者：
相关文件：
收藏者：