Mercurial > hg > smallbox
view toolboxes/AudioInpaintingToolbox/Utils/dictionaries/Gabor_Dictionary.m @ 245:96d17e5dc5d3
Replaced the call to SMALL_ImgDenoise_DL_test_KSVDvsSPAMS with SMALL_ImgDenoise_DL_test_KSVDvsRLSDLAvsTwoStepMOD in Example 2 of SMALLBoxSetup.m.
author | luisf <luis.figueira@eecs.qmul.ac.uk> |
---|---|
date | Wed, 31 Oct 2012 11:53:07 +0000 |
parents | 56d719a5fd31 |
children |
line wrap: on
line source
function D = Gabor_Dictionary(param) % Windowed Gabor dictionary. In this implementation, the dictionary matrix % is the concatenation of a DCT (left part of the matrix) and of a DST % (right part). % Note that one can use this dictionary % - either by constraining the simulaneous selection of cosine and sine % atoms with same frequency in order to implement Gabor atoms; % - or, without any selection constraint, by considering that the % dictionary is not a Gabor dictionary but the concatenation of a DCT and % of a DST. % % Usage: % D = Gabor_Dictionary(param) % % Inputs [and default values]: % - param.N: frame length [256] % - param.redundancyFactor: redundancy factor to adjust the number of % frequencies [1]. The number of atoms in the dictionary equals % param.N*param.redundancyFactor % - param.wd: weigthing window function [@wSine] % % Output: % - Dictionary: D (cosine atoms followed by sine atoms) % % % ------------------- % % Audio Inpainting toolbox % Date: June 28, 2011 % By Valentin Emiya, Amir Adler, Maria Jafari % This code is distributed under the terms of the GNU Public License version 3 (http://www.gnu.org/licenses/gpl.txt). % Check and load default parameters defaultParam.N = 256; defaultParam.redundancyFactor = 1; defaultParam.wd = @wSine; if nargin<1 param = defaultParam; else names = fieldnames(defaultParam); for k=1:length(names) if ~isfield(param,names{k}) || isempty(param.(names{k})) param.(names{k}) = defaultParam.(names{k}); end end end K = param.N*param.redundancyFactor; % number of atoms wd = param.wd(param.N); % weigthing window u = 0:(param.N-1); % time k=0:K/2-1; % frequency D = diag(wd)*[cos(2*pi/K*(u.'+1/2)*(k+1/2)),sin(2*pi/K*(u.'+1/2)*(k+1/2))]; % normalisation D = D*diag(1./sqrt(diag(D'*D))); return