Mercurial > hg > smallbox
changeset 246:cef4500b936f ver_2.1
Merge
| author | luisf <luis.figueira@eecs.qmul.ac.uk> |
|---|---|
| date | Wed, 31 Oct 2012 12:10:13 +0000 |
| parents | 96d17e5dc5d3 (diff) 5c8bcdadb380 (current diff) |
| children | ecce33192fcc |
| files | SMALLboxSetup.m |
| diffstat | 10 files changed, 527 insertions(+), 509 deletions(-) [+] |
line wrap: on
line diff
--- a/.hgtags Tue Sep 04 11:00:36 2012 +0100 +++ b/.hgtags Wed Oct 31 12:10:13 2012 +0000 @@ -5,3 +5,10 @@ 19e0af5709140e163faaf9d8cf4b83a664be1edc release_1.5 30872eb3d1606b89f8d071cfdb7cddb874fab8fd release_1.51 4205744092e6a16951e48acef27fb41e32851758 release_1.9 +15fd4c5be71a70d812fa664817eeeb0954694978 ver_2.0_alpha1 +15fd4c5be71a70d812fa664817eeeb0954694978 ver_2.0_alpha1 +c96880c0c47cf2a17fbd11dacc9aaeaac8ff4ec3 ver_2.0_alpha1 +c96880c0c47cf2a17fbd11dacc9aaeaac8ff4ec3 ver_2.0_alpha1 +5f4e47b78f2b857208951b7570d2d3c5041603ee ver_2.0_alpha1 +71128ec3e532181c002bcbc16b31414f2ec0443e ver_2.0_beta +423de9f24d98de2041f10adba30c36629fcf0c89 ver_2.0
--- a/README.txt Tue Sep 04 11:00:36 2012 +0100 +++ b/README.txt Wed Oct 31 12:10:13 2012 +0000 @@ -1,17 +1,16 @@ -SMALLbox Version 1.0 beta +SMALLbox Version 2.0 -11th May, 2010 +25th May, 2012 -Version 1.0 beta of SMALLbox is the release candidate that is distributed +Version 2.0 of SMALLbox is the release candidate that is distributed for testing and bug fixing purposes. Please send all bugs, requests and suggestions to: -ivan.damnjanovic@elec.qmul.ac.uk - +luis.figueira@soundsoftware.ac.uk --------------------------------------------------------------------------- -Copyright (2010): Ivan Damnjanovic, Matthew Davies +Copyright (2012): Luis Figueira, Ivan Damnjanovic, Matthew Davies Centre for Digital Music, Queen Mary University of London @@ -91,9 +90,7 @@ http://small-project.eu -Contacts: ivan.damnjanovic@elec.qmul.ac.uk - matthew.davies@elec.qmul.ac.uk - +Contact: luis.figueira@soundsoftware.ac.uk This code is in experimental stage; any comments or bug reports are very welcome. More information about using SMALLbox will be includied in release version
--- a/SMALLboxSetup.m Tue Sep 04 11:00:36 2012 +0100 +++ b/SMALLboxSetup.m Wed Oct 31 12:10:13 2012 +0000 @@ -501,7 +501,7 @@ SMALL_solver_test; case{'2'} fprintf('\n Running SMALL Image Denoise problem'); - SMALL_ImgDenoise_DL_test_KSVDvsSPAMS; + SMALL_ImgDenoise_DL_test_KSVDvsRLSDLAvsTwoStepMOD.m; otherwise return; end
--- a/To Do list.m Tue Sep 04 11:00:36 2012 +0100 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,23 +0,0 @@ -CardiacMRI: -- Last thing I was dealing with was temporal fft operator that I have put in opFFTxd.m in util directory. -- Problem setup is now kind of realistic: - - user choose upload cardiac images from dataset - - he can shose the slice number (1-10 - default is 5) - - sigma level (0.05 by default) - - undersampling factor (fold default is 6) - - n image size (256 by default) - - read 3d matrix of n*n cardiac images in szt time points - - adds noise sigma*var(signal) - - do 2d circular fft per image and take random lines as specified by mask - - mask is made to sample n*szt/fold lines in phase encode and time dimensions - - play movies of original, undersampled cardiac images and mask - -SMALLboxSetup.m - Installation script: - -- Add make call for DL/RLS_DLA/private and Problems/private (still need to - shall these be private directories as in KSVD. Files inside are common - common routines that are used all around, so it makes sense to put them - in util directory) - SOLVED: A copy of "ksvd/private" is put into "util/ksvd utils" and istallation - script will run make command in the directory to compile mex files -- \ No newline at end of file
--- a/bugs.m Tue Sep 04 11:00:36 2012 +0100 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,166 +0,0 @@ -% MDP: File "COPYING" is not included in the distribution? - -% I encountered an error while running SMALL_solver_test_Audio : - -%??? Error using ==> -%problems\private\completeOps>@(x)reshape(data.B(x,1),data.signalSize) -%Too many input arguments. -%Error in ==> SMALL_playAudio at 17 -%SMALL.solver.reconstructed = SMALL.Problem.reconstruct(SMALL.solver.solution); -%Error in ==> SMALL_solver_test_Audio at 96 -%SMALL_playAudio(SMALL); - - -% version of Matlab used: 7.9.0.529 - 64 bits - - - -Bugs by Raja Giryes: -Using the tollbox on windows with Matlab R2010b: -bug1: -running SMALL_solver_test_Audio.m -I get: -Starting solver SMALL_cgp... -Solver SMALL_cgp finished task in 251.208410 seconds (cpu time). -Solver SMALL_cgp finished task in 159.905633 seconds (tic-toc time). - -Starting solver SolveBP... -??? Error using ==> eval -Undefined function or method 'SolveBP' for input arguments of type -'function_handle'. - -Error in ==> SMALL_solve at 46 - y = eval([solver.name,'(SparseLab_A, b, n,',solver.param,');']); - -Error in ==> SMALL_solver_test_Audio at 93 -SMALL.solver(i)=SMALL_solve(SMALL.Problem, SMALL.solver(i)); - -running SMALL_solver_test.m -I get: -Starting solver SMALL_cgp... -Solver SMALL_cgp finished task in 2.839218 seconds (cpu time). -Solver SMALL_cgp finished task in 2.821443 seconds (tic-toc time). - -Starting solver SolveOMP... -??? Error using ==> eval -Undefined function or method 'SolveOMP' for input arguments of type -'function_handle'. - -Error in ==> SMALL_solve at 46 - y = eval([solver.name,'(SparseLab_A, b, n,',solver.param,');']); - -Error in ==> SMALL_solver_test at 113 -SMALL.solver(i)=SMALL_solve(SMALL.Problem, SMALL.solver(i)); - - -Bug report from Mehrdad Yaghoobi: -1. The rice mdwt crashed on my matlab session runing on linux 64!!! -2. problem with hg merge using EasyMercurial. - - - - -% MDP: Files "my_dummy_solver.m" etc should include the boiler plate function -% header etc. - - - - -%%%%%%%%%%%%%%%%% -% bug by Valentin -%%%%%%%%%%%%%%%%% -% In SMALLboxSetup.m -% On Windows XP, 32 bits, Matlab R2010a -% -% > SMALLboxSetup -% ... -% Compiling the Rice Wavelet Toolbox MEX interfaces...Error mdwt.c: 39 undeclared identifier `intptr_t' -% Warning mdwt.c: 39 Statement has no effect -% Error mdwt.c: 39 syntax error; found `m' expecting `;' -% Error mdwt.c: 39 undeclared identifier `m' -% Error mdwt.c: 39 undeclared identifier `n' -% Error mdwt.c: 39 undeclared identifier `h_col' -% Error mdwt.c: 39 undeclared identifier `h_row' -% Error mdwt.c: 39 undeclared identifier `lh' -% Error mdwt.c: 39 undeclared identifier `L' -% Error mdwt.c: 39 undeclared identifier `i' -% Error mdwt.c: 39 undeclared identifier `po2' -% Error mdwt.c: 39 undeclared identifier `j' -% Warning mdwt.c: 39 Statement has no effect -% Error mdwt.c: 40 illegal statement termination -% Error mdwt.c: 40 skipping `double' -% Error mdwt.c: 40 undeclared identifier `mtest' -% Error mdwt.c: 40 undeclared identifier `ntest' -% Warning mdwt.c: 40 Statement has no effect -% Error mdwt.c: 62 operands of * have illegal types `int' and `pointer to double' -% Warning mdwt.c: 39 possible usage of po2 before definition -% Warning mdwt.c: 39 possible usage of intptr_t before definition -% Warning mdwt.c: 38 local `pointer to double Lf' is not referenced -% Error mdwt.c: 106 missing parameter type -% Error mdwt.c: 106 syntax error; found `m' expecting `)' -% Error mdwt.c: 106 skipping `m' `,' `intptr_t' `n' `,' -% Error mdwt.c: 106 syntax error; found `double' expecting `{' -% Error mdwt.c: 106 too many errors -% -% C:\PROGRA~1\MATLAB\R2010A\BIN\MEX.PL: Error: Compile of 'mdwt.c' failed. -% -% Warning: Could not compile Rice Wavelet Toolbox MEX interfaces.\n -% > In SMALLboxSetup at 365 -% -% ****************************************************************** -% -% Initialising SMALLbox Examples Setup -% ****************************************************************** -% -% SMALLbox Installation Complete! -% -% For more information on the installed toolboxes see -% -% Sparco: http://www.cs.ubc.ca/labs/scl/sparco/ -% -% SPGL1: http://www.cs.ubc.ca/labs/spgl1/?n=HomePage -% -% SparseLab: http://sparselab.stanford.edu/ (PLEASE REGISTER SPARSELAB!) -% -% Sparsify: http://www.see.ed.ac.uk/~tblumens/sparsify/sparsify.html -% -% GPSR: http://www.lx.it.pt/~mtf/GPSR/ -% -% OMPbox and KSVDBox: http://www.cs.technion.ac.il/~ronrubin/ -% -% -% Would you like to run a demo: ([y]/n)? -% -% 1 to run SMALL_solver_test -% 2 to run Dictionary Learning for Image Denoising demo -% q to quit: 1 -% -% Running SMALL_solver_test problem -% -% Example test of SMALL solvers against their counterparts on Sparco problems. -% -% ??? Undefined function or method 'mdwt' for input arguments of type -% 'double'. -% -% Error in ==> opWavelet>opWavelet_intrnl at 55 -% [y,l] = mdwt(reshape(x,[m,n]), h, levels); -% -% Error in ==> -% opWavelet>@(x,mode)opWavelet_intrnl(m,n,family,filter,levels,type,h,x,mode) -% at 37 -% op = @(x,mode) -% opWavelet_intrnl(m,n,family,filter,levels,type,h,x,mode); -% -% Error in ==> prob006 at 67 -% data.x0 = data.op.Daubechies(data.signal,2); -% -% Error in ==> generateProblem at 101 -% [data] = eval(sprintf('prob%03d(varargin{:});',index)); -% -% Error in ==> SMALL_solver_test at 61 -% SMALL.Problem = generateProblem(6, 'P', 6, 'm', 270,'n',1024, -% 'show'); -% -% Error in ==> SMALLboxSetup at 416 -% SMALL_solver_test; -% \ No newline at end of file
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/examples/Image Denoising/SMALL_ImgDenoise_DL_test_KSVDvsSKSVD.m Wed Oct 31 12:10:13 2012 +0000 @@ -0,0 +1,204 @@ +%% Dictionary Learning for Image Denoising - KSVD vs KSVDS vs SPAMS +% +% *WARNING!* You should have SPAMS in your search path in order for this +% script to work.Due to licensing issues SPAMS can not be automatically +% provided in SMALLbox (http://www.di.ens.fr/willow/SPAMS/downloads.html). +% +% This file contains an example of how SMALLbox can be used to test different +% dictionary learning techniques in Image Denoising problem. +% It calls generateImageDenoiseProblem that will let you to choose image, +% add noise and use noisy image to generate training set for dictionary +% learning. +% Two dictionary learning techniques were compared: +% - KSVD - M. Elad, R. Rubinstein, and M. Zibulevsky, "Efficient +% Implementation of the K-SVD Algorithm using Batch Orthogonal +% Matching Pursuit", Technical Report - CS, Technion, April 2008. +% - KSVDS - R. Rubinstein, M. Zibulevsky, and M. Elad, "Learning Sparse +% Dictionaries for Sparse Signal Approximation", Technical +% Report - CS, Technion, June 2009. +% + +% +% Centre for Digital Music, Queen Mary, University of London. +% This file copyright 2009 Ivan Damnjanovic. +% +% This program is free software; you can redistribute it and/or +% modify it under the terms of the GNU General Public License as +% published by the Free Software Foundation; either version 2 of the +% License, or (at your option) any later version. See the file +% COPYING included with this distribution for more information. +% +%% + +clear; + +% If you want to load the image outside of generateImageDenoiseProblem +% function uncomment following lines. This can be useful if you want to +% denoise more then one image for example. + +% TMPpath=pwd; +% FS=filesep; +% [pathstr1, name, ext] = fileparts(which('SMALLboxSetup.m')); +% cd([pathstr1,FS,'data',FS,'images']); +% [filename,pathname] = uigetfile({'*.png;'},'Select a file containin pre-calculated notes'); +% [pathstr, name, ext] = fileparts(filename); +% test_image = imread(filename); +% test_image = double(test_image); +% cd(TMPpath); +% SMALL.Problem.name=name; + + +% Defining Image Denoising Problem as Dictionary Learning +% Problem. As an input we set the number of training patches. + +SMALL.Problem = generateImageDenoiseProblem('', 40000); + + +%% +% Use KSVD Dictionary Learning Algorithm to Learn overcomplete dictionary + +% Initialising Dictionary structure +% Setting Dictionary structure fields (toolbox, name, param, D and time) +% to zero values + +SMALL.DL(1)=SMALL_init_DL(); + +% Defining the parameters needed for dictionary learning + +SMALL.DL(1).toolbox = 'KSVD'; +SMALL.DL(1).name = 'ksvd'; + +% Defining the parameters for KSVD +% In this example we are learning 256 atoms in 20 iterations, so that +% every patch in the training set can be represented with target error in +% L2-norm (EData) +% Type help ksvd in MATLAB prompt for more options. + +Edata=sqrt(prod(SMALL.Problem.blocksize)) * SMALL.Problem.sigma * SMALL.Problem.gain; +maxatoms = floor(prod(SMALL.Problem.blocksize)/2); + +SMALL.DL(1).param=struct(... + 'Edata', Edata,... + 'initdict', SMALL.Problem.initdict,... + 'dictsize', SMALL.Problem.p,... + 'iternum', 20,... + 'memusage', 'high'); + +% Learn the dictionary + +SMALL.DL(1) = SMALL_learn(SMALL.Problem, SMALL.DL(1)); + +% Set SMALL.Problem.A dictionary +% (backward compatiblity with SPARCO: solver structure communicate +% only with Problem structure, ie no direct communication between DL and +% solver structures) + +SMALL.Problem.A = SMALL.DL(1).D; +SMALL.Problem.reconstruct = @(x) ImageDenoise_reconstruct(x, SMALL.Problem); + +%% +% Initialising solver structure +% Setting solver structure fields (toolbox, name, param, solution, +% reconstructed and time) to zero values + +SMALL.solver(1)=SMALL_init_solver; + +% Defining the parameters needed for image denoising + +SMALL.solver(1).toolbox='ompbox'; +SMALL.solver(1).name='omp2'; +SMALL.solver(1).param=struct(... + 'epsilon',Edata,... + 'maxatoms', maxatoms); + +% Denoising the image - find the sparse solution in the learned +% dictionary for all patches in the image and the end it uses +% reconstruction function to reconstruct the patches and put them into a +% denoised image + +SMALL.solver(1)=SMALL_solve(SMALL.Problem, SMALL.solver(1)); + +% Show PSNR after reconstruction + +SMALL.solver(1).reconstructed.psnr + +%% +% Use KSVDS Dictionary Learning Algorithm to denoise image + +% Initialising solver structure +% Setting solver structure fields (toolbox, name, param, solution, +% reconstructed and time) to zero values + +SMALL.DL(2)=SMALL_init_DL(); + +% Defining the parameters needed for dictionary learning + +SMALL.DL(2).toolbox = 'KSVDS'; +SMALL.DL(2).name = 'ksvds'; + +% Defining the parameters for KSVDS +% In this example we are learning 256 atoms in 20 iterations, so that +% every patch in the training set can be represented with target error in +% L2-norm (EDataS). We also impose "double sparsity" - dictionary itself +% has to be sparse in the given base dictionary (Tdict - number of +% nonzero elements per atom). +% Type help ksvds in MATLAB prompt for more options. + +EdataS=sqrt(prod(SMALL.Problem.blocksize)) * SMALL.Problem.sigma * SMALL.Problem.gain; +SMALL.DL(2).param=struct(... + 'Edata', EdataS, ... + 'Tdict', 6,... + 'stepsize', 1,... + 'dictsize', SMALL.Problem.p,... + 'iternum', 20,... + 'memusage', 'high'); +SMALL.DL(2).param.initA = speye(SMALL.Problem.p); +SMALL.DL(2).param.basedict{1} = odctdict(8,16); +SMALL.DL(2).param.basedict{2} = odctdict(8,16); + +% Learn the dictionary + +SMALL.DL(2) = SMALL_learn(SMALL.Problem, SMALL.DL(2)); + +% Set SMALL.Problem.A dictionary and SMALL.Problem.basedictionary +% (backward compatiblity with SPARCO: solver structure communicate +% only with Problem structure, ie no direct communication between DL and +% solver structures) + +SMALL.Problem.A = SMALL.DL(2).D; +SMALL.Problem.basedict{1} = SMALL.DL(2).param.basedict{1}; +SMALL.Problem.basedict{2} = SMALL.DL(2).param.basedict{2}; + +% Setting up reconstruction function + +SparseDict=1; +SMALL.Problem.reconstruct = @(x) ImageDenoise_reconstruct(x, SMALL.Problem, SparseDict); + +% Initialising solver structure +% Setting solver structure fields (toolbox, name, param, solution, +% reconstructed and time) to zero values + +SMALL.solver(2)=SMALL_init_solver; + +% Defining the parameters needed for image denoising + +SMALL.solver(2).toolbox='ompsbox'; +SMALL.solver(2).name='omps2'; +SMALL.solver(2).param=struct(... + 'epsilon',Edata,... + 'maxatoms', maxatoms); + +% Denoising the image - find the sparse solution in the learned +% dictionary for all patches in the image and the end it uses +% reconstruction function to reconstruct the patches and put them into a +% denoised image + +SMALL.solver(2)=SMALL_solve(SMALL.Problem, SMALL.solver(2)); + + +%% +% Plot results and save midi files + +% show results % + +SMALL_ImgDeNoiseResult(SMALL);
--- a/examples/Image Denoising/SMALL_ImgDenoise_DL_test_TwoStep_KSVD_MOD_OLS_Mailhe.m Tue Sep 04 11:00:36 2012 +0100 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,309 +0,0 @@ -%% Dictionary Learning for Image Denoising - KSVD vs Recursive Least Squares -% -% This file contains an example of how SMALLbox can be used to test different -% dictionary learning techniques in Image Denoising problem. -% It calls generateImageDenoiseProblem that will let you to choose image, -% add noise and use noisy image to generate training set for dictionary -% learning. -% Two dictionary learning techniques were compared: -% - KSVD - M. Elad, R. Rubinstein, and M. Zibulevsky, "Efficient -% Implementation of the K-SVD Algorithm using Batch Orthogonal -% Matching Pursuit", Technical Report - CS, Technion, April 2008. -% - RLS-DLA - Skretting, K.; Engan, K.; , "Recursive Least Squares -% Dictionary Learning Algorithm," Signal Processing, IEEE Transactions on, -% vol.58, no.4, pp.2121-2130, April 2010 -% - - -% Centre for Digital Music, Queen Mary, University of London. -% This file copyright 2011 Ivan Damnjanovic. -% -% This program is free software; you can redistribute it and/or -% modify it under the terms of the GNU General Public License as -% published by the Free Software Foundation; either version 2 of the -% License, or (at your option) any later version. See the file -% COPYING included with this distribution for more information. -% -%% - - - -% If you want to load the image outside of generateImageDenoiseProblem -% function uncomment following lines. This can be useful if you want to -% denoise more then one image for example. -% Here we are loading test_image.mat that contains structure with 5 images : lena, -% barbara,boat, house and peppers. -clear; -TMPpath=pwd; -FS=filesep; -[pathstr1, name, ext] = fileparts(which('SMALLboxSetup.m')); -cd([pathstr1,FS,'data',FS,'images']); -load('test_image.mat'); -cd(TMPpath); - -% Deffining the noise levels that we want to test - -noise_level=[10 20 25 50 100]; - -% Here we loop through different noise levels and images - -for noise_ind=2:2 -for im_num=2:2 - -% Defining Image Denoising Problem as Dictionary Learning -% Problem. As an input we set the number of training patches. - -SMALL.Problem = generateImageDenoiseProblem(test_image(im_num).i, 40000, '',256, noise_level(noise_ind)); -SMALL.Problem.name=int2str(im_num); - -Edata=sqrt(prod(SMALL.Problem.blocksize)) * SMALL.Problem.sigma * SMALL.Problem.gain; -maxatoms = floor(prod(SMALL.Problem.blocksize)/2); - - -%% -% Use KSVD Dictionary Learning Algorithm to Learn overcomplete dictionary -% Boris Mailhe ksvd update implentation omp is the same as with Rubinstein -% implementation - - -% Initialising solver structure -% Setting solver structure fields (toolbox, name, param, solution, -% reconstructed and time) to zero values - -SMALL.solver(1)=SMALL_init_solver; - -% Defining the parameters needed for image denoising - -SMALL.solver(1).toolbox='ompbox'; -SMALL.solver(1).name='omp2'; -SMALL.solver(1).param=struct(... - 'epsilon',Edata,... - 'maxatoms', maxatoms); - -% Initialising Dictionary structure -% Setting Dictionary structure fields (toolbox, name, param, D and time) -% to zero values - -SMALL.DL(1)=SMALL_init_DL('TwoStepDL', 'KSVD', '', 1); - - -% Defining the parameters for KSVD -% In this example we are learning 256 atoms in 20 iterations, so that -% every patch in the training set can be represented with target error in -% L2-norm (EData) -% Type help ksvd in MATLAB prompt for more options. - - -SMALL.DL(1).param=struct(... - 'solver', SMALL.solver(1),... - 'initdict', SMALL.Problem.initdict,... - 'dictsize', SMALL.Problem.p,... - 'iternum', 20,... - 'show_dict', 1); - -% Learn the dictionary - -SMALL.DL(1) = SMALL_learn(SMALL.Problem, SMALL.DL(1)); - -% Set SMALL.Problem.A dictionary -% (backward compatiblity with SPARCO: solver structure communicate -% only with Problem structure, ie no direct communication between DL and -% solver structures) - -SMALL.Problem.A = SMALL.DL(1).D; -SMALL.Problem.reconstruct = @(x) ImageDenoise_reconstruct(x, SMALL.Problem); - -% Denoising the image - find the sparse solution in the learned -% dictionary for all patches in the image and the end it uses -% reconstruction function to reconstruct the patches and put them into a -% denoised image - -SMALL.solver(1)=SMALL_solve(SMALL.Problem, SMALL.solver(1)); - -%% -% Use MOD Dictionary Learning Algorithm to Learn overcomplete dictionary -% Boris Mailhe MOD update implentation omp is the Ron Rubinstein -% implementation - - -% Initialising solver structure -% Setting solver structure fields (toolbox, name, param, solution, -% reconstructed and time) to zero values - -SMALL.solver(2)=SMALL_init_solver; - -% Defining the parameters needed for image denoising - -SMALL.solver(2).toolbox='ompbox'; -SMALL.solver(2).name='omp2'; -SMALL.solver(2).param=struct(... - 'epsilon',Edata,... - 'maxatoms', maxatoms); - -% Initialising Dictionary structure -% Setting Dictionary structure fields (toolbox, name, param, D and time) -% to zero values - -SMALL.DL(2)=SMALL_init_DL('TwoStepDL', 'MOD', '', 1); - - -% Defining the parameters for MOD -% In this example we are learning 256 atoms in 20 iterations, so that -% every patch in the training set can be represented with target error in -% L2-norm (EData) -% Type help ksvd in MATLAB prompt for more options - -SMALL.DL(2).param=struct(... - 'solver', SMALL.solver(2),... - 'initdict', SMALL.Problem.initdict,... - 'dictsize', SMALL.Problem.p,... - 'iternum', 20,... - 'show_dict', 1); - -% Learn the dictionary - -SMALL.DL(2) = SMALL_learn(SMALL.Problem, SMALL.DL(2)); - -% Set SMALL.Problem.A dictionary -% (backward compatiblity with SPARCO: solver structure communicate -% only with Problem structure, ie no direct communication between DL and -% solver structures) - -SMALL.Problem.A = SMALL.DL(2).D; -SMALL.Problem.reconstruct = @(x) ImageDenoise_reconstruct(x, SMALL.Problem); - -% Denoising the image - find the sparse solution in the learned -% dictionary for all patches in the image and the end it uses -% reconstruction function to reconstruct the patches and put them into a -% denoised image - -SMALL.solver(2)=SMALL_solve(SMALL.Problem, SMALL.solver(2)); -%% -% Use OLS Dictionary Learning Algorithm to Learn overcomplete dictionary -% Boris Mailhe ksvd update implentation omp is the Ron Rubinstein -% implementation - - -% Initialising solver structure -% Setting solver structure fields (toolbox, name, param, solution, -% reconstructed and time) to zero values - -SMALL.solver(3)=SMALL_init_solver; - -% Defining the parameters needed for image denoising - -SMALL.solver(3).toolbox='ompbox'; -SMALL.solver(3).name='omp2'; -SMALL.solver(3).param=struct(... - 'epsilon',Edata,... - 'maxatoms', maxatoms); - -% Initialising Dictionary structure -% Setting Dictionary structure fields (toolbox, name, param, D and time) -% to zero values - -SMALL.DL(3)=SMALL_init_DL('TwoStepDL', 'ols', '', 1); - - -% Defining the parameters for KSVD -% In this example we are learning 256 atoms in 20 iterations, so that -% every patch in the training set can be represented with target error in -% L2-norm (EData) -% Type help ksvd in MATLAB prompt for more options. - - -SMALL.DL(3).param=struct(... - 'solver', SMALL.solver(3),... - 'initdict', SMALL.Problem.initdict,... - 'dictsize', SMALL.Problem.p,... - 'iternum', 20,... - 'learningRate', 0.1,... - 'show_dict', 1); - -% Learn the dictionary - -SMALL.DL(3) = SMALL_learn(SMALL.Problem, SMALL.DL(3)); - -% Set SMALL.Problem.A dictionary -% (backward compatiblity with SPARCO: solver structure communicate -% only with Problem structure, ie no direct communication between DL and -% solver structures) - -SMALL.Problem.A = SMALL.DL(3).D; -SMALL.Problem.reconstruct = @(x) ImageDenoise_reconstruct(x, SMALL.Problem); - -% Denoising the image - find the sparse solution in the learned -% dictionary for all patches in the image and the end it uses -% reconstruction function to reconstruct the patches and put them into a -% denoised image - -SMALL.solver(3)=SMALL_solve(SMALL.Problem, SMALL.solver(3)); -%% -% Use Mailhe Dictionary Learning Algorithm to Learn overcomplete dictionary -% Boris Mailhe ksvd update implentation omp is the Ron Rubinstein -% implementation - - -% Initialising solver structure -% Setting solver structure fields (toolbox, name, param, solution, -% reconstructed and time) to zero values - -SMALL.solver(4)=SMALL_init_solver; - -% Defining the parameters needed for image denoising - -SMALL.solver(4).toolbox='ompbox'; -SMALL.solver(4).name='omp2'; -SMALL.solver(4).param=struct(... - 'epsilon',Edata,... - 'maxatoms', maxatoms); - -% Initialising Dictionary structure -% Setting Dictionary structure fields (toolbox, name, param, D and time) -% to zero values - -SMALL.DL(4)=SMALL_init_DL('TwoStepDL', 'opt', '', 1); - - -% Defining the parameters for KSVD -% In this example we are learning 256 atoms in 20 iterations, so that -% every patch in the training set can be represented with target error in -% L2-norm (EData) -% Type help ksvd in MATLAB prompt for more options. - - -SMALL.DL(4).param=struct(... - 'solver', SMALL.solver(4),... - 'initdict', SMALL.Problem.initdict,... - 'dictsize', SMALL.Problem.p,... - 'iternum', 20,... - 'learningRate', 2,... - 'show_dict', 1); - -% Learn the dictionary - -SMALL.DL(4) = SMALL_learn(SMALL.Problem, SMALL.DL(4)); - -% Set SMALL.Problem.A dictionary -% (backward compatiblity with SPARCO: solver structure communicate -% only with Problem structure, ie no direct communication between DL and -% solver structures) - -SMALL.Problem.A = SMALL.DL(4).D; -SMALL.Problem.reconstruct = @(x) ImageDenoise_reconstruct(x, SMALL.Problem); - -% Denoising the image - find the sparse solution in the learned -% dictionary for all patches in the image and the end it uses -% reconstruction function to reconstruct the patches and put them into a -% denoised image - -SMALL.solver(4)=SMALL_solve(SMALL.Problem, SMALL.solver(4)); - -%% show results %% - -SMALL_ImgDeNoiseResult(SMALL); - -%clear SMALL; -end -end -
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/examples/Image Denoising/SMALL_ImgDenoise_DL_test_TwoStep_KSVD_MOD_OLS_OPT.m Wed Oct 31 12:10:13 2012 +0000 @@ -0,0 +1,309 @@ +%% Dictionary Learning for Image Denoising - KSVD vs Recursive Least Squares +% +% This file contains an example of how SMALLbox can be used to test different +% dictionary learning techniques in Image Denoising problem. +% It calls generateImageDenoiseProblem that will let you to choose image, +% add noise and use noisy image to generate training set for dictionary +% learning. +% Two dictionary learning techniques were compared: +% - KSVD - M. Elad, R. Rubinstein, and M. Zibulevsky, "Efficient +% Implementation of the K-SVD Algorithm using Batch Orthogonal +% Matching Pursuit", Technical Report - CS, Technion, April 2008. +% - RLS-DLA - Skretting, K.; Engan, K.; , "Recursive Least Squares +% Dictionary Learning Algorithm," Signal Processing, IEEE Transactions on, +% vol.58, no.4, pp.2121-2130, April 2010 +% + + +% Centre for Digital Music, Queen Mary, University of London. +% This file copyright 2011 Ivan Damnjanovic. +% +% This program is free software; you can redistribute it and/or +% modify it under the terms of the GNU General Public License as +% published by the Free Software Foundation; either version 2 of the +% License, or (at your option) any later version. See the file +% COPYING included with this distribution for more information. +% +%% + + + +% If you want to load the image outside of generateImageDenoiseProblem +% function uncomment following lines. This can be useful if you want to +% denoise more then one image for example. +% Here we are loading test_image.mat that contains structure with 5 images : lena, +% barbara,boat, house and peppers. +clear; +TMPpath=pwd; +FS=filesep; +[pathstr1, name, ext] = fileparts(which('SMALLboxSetup.m')); +cd([pathstr1,FS,'data',FS,'images']); +load('test_image.mat'); +cd(TMPpath); + +% Deffining the noise levels that we want to test + +noise_level=[10 20 25 50 100]; + +% Here we loop through different noise levels and images + +for noise_ind=2:2 +for im_num=2:2 + +% Defining Image Denoising Problem as Dictionary Learning +% Problem. As an input we set the number of training patches. + +SMALL.Problem = generateImageDenoiseProblem(test_image(im_num).i, 40000, '',256, noise_level(noise_ind)); +SMALL.Problem.name=int2str(im_num); + +Edata=sqrt(prod(SMALL.Problem.blocksize)) * SMALL.Problem.sigma * SMALL.Problem.gain; +maxatoms = floor(prod(SMALL.Problem.blocksize)/2); + + +%% +% Use KSVD Dictionary Learning Algorithm to Learn overcomplete dictionary +% Boris Mailhe ksvd update implentation omp is the same as with Rubinstein +% implementation + + +% Initialising solver structure +% Setting solver structure fields (toolbox, name, param, solution, +% reconstructed and time) to zero values + +SMALL.solver(1)=SMALL_init_solver; + +% Defining the parameters needed for image denoising + +SMALL.solver(1).toolbox='ompbox'; +SMALL.solver(1).name='omp2'; +SMALL.solver(1).param=struct(... + 'epsilon',Edata,... + 'maxatoms', maxatoms); + +% Initialising Dictionary structure +% Setting Dictionary structure fields (toolbox, name, param, D and time) +% to zero values + +SMALL.DL(1)=SMALL_init_DL('TwoStepDL', 'KSVD', '', 1); + + +% Defining the parameters for KSVD +% In this example we are learning 256 atoms in 20 iterations, so that +% every patch in the training set can be represented with target error in +% L2-norm (EData) +% Type help ksvd in MATLAB prompt for more options. + + +SMALL.DL(1).param=struct(... + 'solver', SMALL.solver(1),... + 'initdict', SMALL.Problem.initdict,... + 'dictsize', SMALL.Problem.p,... + 'iternum', 20,... + 'show_dict', 1); + +% Learn the dictionary + +SMALL.DL(1) = SMALL_learn(SMALL.Problem, SMALL.DL(1)); + +% Set SMALL.Problem.A dictionary +% (backward compatiblity with SPARCO: solver structure communicate +% only with Problem structure, ie no direct communication between DL and +% solver structures) + +SMALL.Problem.A = SMALL.DL(1).D; +SMALL.Problem.reconstruct = @(x) ImageDenoise_reconstruct(x, SMALL.Problem); + +% Denoising the image - find the sparse solution in the learned +% dictionary for all patches in the image and the end it uses +% reconstruction function to reconstruct the patches and put them into a +% denoised image + +SMALL.solver(1)=SMALL_solve(SMALL.Problem, SMALL.solver(1)); + +%% +% Use MOD Dictionary Learning Algorithm to Learn overcomplete dictionary +% Boris Mailhe MOD update implentation omp is the Ron Rubinstein +% implementation + + +% Initialising solver structure +% Setting solver structure fields (toolbox, name, param, solution, +% reconstructed and time) to zero values + +SMALL.solver(2)=SMALL_init_solver; + +% Defining the parameters needed for image denoising + +SMALL.solver(2).toolbox='ompbox'; +SMALL.solver(2).name='omp2'; +SMALL.solver(2).param=struct(... + 'epsilon',Edata,... + 'maxatoms', maxatoms); + +% Initialising Dictionary structure +% Setting Dictionary structure fields (toolbox, name, param, D and time) +% to zero values + +SMALL.DL(2)=SMALL_init_DL('TwoStepDL', 'MOD', '', 1); + + +% Defining the parameters for MOD +% In this example we are learning 256 atoms in 20 iterations, so that +% every patch in the training set can be represented with target error in +% L2-norm (EData) +% Type help ksvd in MATLAB prompt for more options + +SMALL.DL(2).param=struct(... + 'solver', SMALL.solver(2),... + 'initdict', SMALL.Problem.initdict,... + 'dictsize', SMALL.Problem.p,... + 'iternum', 20,... + 'show_dict', 1); + +% Learn the dictionary + +SMALL.DL(2) = SMALL_learn(SMALL.Problem, SMALL.DL(2)); + +% Set SMALL.Problem.A dictionary +% (backward compatiblity with SPARCO: solver structure communicate +% only with Problem structure, ie no direct communication between DL and +% solver structures) + +SMALL.Problem.A = SMALL.DL(2).D; +SMALL.Problem.reconstruct = @(x) ImageDenoise_reconstruct(x, SMALL.Problem); + +% Denoising the image - find the sparse solution in the learned +% dictionary for all patches in the image and the end it uses +% reconstruction function to reconstruct the patches and put them into a +% denoised image + +SMALL.solver(2)=SMALL_solve(SMALL.Problem, SMALL.solver(2)); +%% +% Use OLS Dictionary Learning Algorithm to Learn overcomplete dictionary +% Boris Mailhe ksvd update implentation omp is the Ron Rubinstein +% implementation + + +% Initialising solver structure +% Setting solver structure fields (toolbox, name, param, solution, +% reconstructed and time) to zero values + +SMALL.solver(3)=SMALL_init_solver; + +% Defining the parameters needed for image denoising + +SMALL.solver(3).toolbox='ompbox'; +SMALL.solver(3).name='omp2'; +SMALL.solver(3).param=struct(... + 'epsilon',Edata,... + 'maxatoms', maxatoms); + +% Initialising Dictionary structure +% Setting Dictionary structure fields (toolbox, name, param, D and time) +% to zero values + +SMALL.DL(3)=SMALL_init_DL('TwoStepDL', 'ols', '', 1); + + +% Defining the parameters for KSVD +% In this example we are learning 256 atoms in 20 iterations, so that +% every patch in the training set can be represented with target error in +% L2-norm (EData) +% Type help ksvd in MATLAB prompt for more options. + + +SMALL.DL(3).param=struct(... + 'solver', SMALL.solver(3),... + 'initdict', SMALL.Problem.initdict,... + 'dictsize', SMALL.Problem.p,... + 'iternum', 20,... + 'learningRate', 0.1,... + 'show_dict', 1); + +% Learn the dictionary + +SMALL.DL(3) = SMALL_learn(SMALL.Problem, SMALL.DL(3)); + +% Set SMALL.Problem.A dictionary +% (backward compatiblity with SPARCO: solver structure communicate +% only with Problem structure, ie no direct communication between DL and +% solver structures) + +SMALL.Problem.A = SMALL.DL(3).D; +SMALL.Problem.reconstruct = @(x) ImageDenoise_reconstruct(x, SMALL.Problem); + +% Denoising the image - find the sparse solution in the learned +% dictionary for all patches in the image and the end it uses +% reconstruction function to reconstruct the patches and put them into a +% denoised image + +SMALL.solver(3)=SMALL_solve(SMALL.Problem, SMALL.solver(3)); +%% +% Use Mailhe Dictionary Learning Algorithm to Learn overcomplete dictionary +% Boris Mailhe ksvd update implentation omp is the Ron Rubinstein +% implementation + + +% Initialising solver structure +% Setting solver structure fields (toolbox, name, param, solution, +% reconstructed and time) to zero values + +SMALL.solver(4)=SMALL_init_solver; + +% Defining the parameters needed for image denoising + +SMALL.solver(4).toolbox='ompbox'; +SMALL.solver(4).name='omp2'; +SMALL.solver(4).param=struct(... + 'epsilon',Edata,... + 'maxatoms', maxatoms); + +% Initialising Dictionary structure +% Setting Dictionary structure fields (toolbox, name, param, D and time) +% to zero values + +SMALL.DL(4)=SMALL_init_DL('TwoStepDL', 'opt', '', 1); + + +% Defining the parameters for KSVD +% In this example we are learning 256 atoms in 20 iterations, so that +% every patch in the training set can be represented with target error in +% L2-norm (EData) +% Type help ksvd in MATLAB prompt for more options. + + +SMALL.DL(4).param=struct(... + 'solver', SMALL.solver(4),... + 'initdict', SMALL.Problem.initdict,... + 'dictsize', SMALL.Problem.p,... + 'iternum', 20,... + 'learningRate', 2,... + 'show_dict', 1); + +% Learn the dictionary + +SMALL.DL(4) = SMALL_learn(SMALL.Problem, SMALL.DL(4)); + +% Set SMALL.Problem.A dictionary +% (backward compatiblity with SPARCO: solver structure communicate +% only with Problem structure, ie no direct communication between DL and +% solver structures) + +SMALL.Problem.A = SMALL.DL(4).D; +SMALL.Problem.reconstruct = @(x) ImageDenoise_reconstruct(x, SMALL.Problem); + +% Denoising the image - find the sparse solution in the learned +% dictionary for all patches in the image and the end it uses +% reconstruction function to reconstruct the patches and put them into a +% denoised image + +SMALL.solver(4)=SMALL_solve(SMALL.Problem, SMALL.solver(4)); + +%% show results %% + +SMALL_ImgDeNoiseResult(SMALL); + +%clear SMALL; +end +end +