Merge pull request #49 from vais-ral/dTV

dTV regulariser (2D/3D CPU/GPU)

2 files changed, 71 insertions, 11 deletions

diff --git a/Wrappers/Matlab/demos/demoMatlab_3Ddenoise.m b/Wrappers/Matlab/demos/demoMatlab_3Ddenoise.m
index 71082e7..dc49d9c 100644
--- a/Wrappers/Matlab/demos/demoMatlab_3Ddenoise.m
+++ b/Wrappers/Matlab/demos/demoMatlab_3Ddenoise.m
@@ -1,5 +1,6 @@
 % Volume (3D) denoising demo using CCPi-RGL
-
+clear
+close all
 addpath('../mex_compile/installed');
 addpath('../../../data/');
 
@@ -14,31 +15,65 @@ vol3D(vol3D < 0) = 0;
 figure; imshow(vol3D(:,:,15), [0 1]); title('Noisy image');
 
 %%
-fprintf('Denoise using ROF-TV model (CPU) \n');
+fprintf('Denoise a volume using the ROF-TV model (CPU) \n');
 lambda_rof = 0.03; % regularisation parameter
 tau_rof = 0.0025; % time-marching constant 
 iter_rof = 300; % number of ROF iterations
 tic; u_rof = ROF_TV(single(vol3D), lambda_rof, iter_rof, tau_rof); toc; 
 figure; imshow(u_rof(:,:,15), [0 1]); title('ROF-TV denoised volume (CPU)');
 %%
-% fprintf('Denoise using ROF-TV model (GPU) \n');
+% fprintf('Denoise a volume using the ROF-TV model (GPU) \n');
 % lambda_rof = 0.03; % regularisation parameter
 % tau_rof = 0.0025; % time-marching constant 
 % iter_rof = 300; % number of ROF iterations
 % tic; u_rofG = ROF_TV_GPU(single(vol3D), lambda_rof, iter_rof, tau_rof); toc;
 % figure; imshow(u_rofG(:,:,15), [0 1]); title('ROF-TV denoised volume (GPU)');
 %%
-fprintf('Denoise using FGP-TV model (CPU) \n');
+fprintf('Denoise a volume using the FGP-TV model (CPU) \n');
 lambda_fgp = 0.03; % regularisation parameter
 iter_fgp = 300; % number of FGP iterations
 epsil_tol =  1.0e-05; % tolerance
 tic; u_fgp = FGP_TV(single(vol3D), lambda_fgp, iter_fgp, epsil_tol); toc; 
 figure; imshow(u_fgp(:,:,15), [0 1]); title('FGP-TV denoised volume (CPU)');
 %%
-% fprintf('Denoise using FGP-TV model (GPU) \n');
+% fprintf('Denoise a volume using the FGP-TV model (GPU) \n');
 % lambda_fgp = 0.03; % regularisation parameter
 % iter_fgp = 300; % number of FGP iterations
 % epsil_tol =  1.0e-05; % tolerance
 % tic; u_fgpG = FGP_TV_GPU(single(vol3D), lambda_fgp, iter_fgp, epsil_tol); toc; 
 % figure; imshow(u_fgpG(:,:,15), [0 1]); title('FGP-TV denoised volume (GPU)');
 %%
+fprintf('Denoise a volume using the FGP-dTV model (CPU) \n');
+
+% create another volume (reference) with slightly less amount of noise
+vol3D_ref = zeros(N,N,slices, 'single');
+for i = 1:slices
+vol3D_ref(:,:,i) = Im + .01*randn(size(Im)); 
+end
+vol3D_ref(vol3D_ref < 0) = 0;
+% vol3D_ref = zeros(size(Im),'single'); % pass zero reference (dTV -> TV)
+
+lambda_fgp = 0.03; % regularisation parameter
+iter_fgp = 300; % number of FGP iterations
+epsil_tol =  1.0e-05; % tolerance
+eta =  0.2; % Reference image gradient smoothing constant
+tic; u_fgp_dtv = FGP_dTV(single(vol3D), single(vol3D_ref), lambda_fgp, iter_fgp, epsil_tol, eta); toc; 
+figure; imshow(u_fgp_dtv(:,:,15), [0 1]); title('FGP-dTV denoised volume (CPU)');
+%%
+fprintf('Denoise a volume using the FGP-dTV model (GPU) \n');
+
+% create another volume (reference) with slightly less amount of noise
+vol3D_ref = zeros(N,N,slices, 'single');
+for i = 1:slices
+vol3D_ref(:,:,i) = Im + .01*randn(size(Im)); 
+end
+vol3D_ref(vol3D_ref < 0) = 0;
+% vol3D_ref = zeros(size(Im),'single'); % pass zero reference (dTV -> TV)
+
+lambda_fgp = 0.03; % regularisation parameter
+iter_fgp = 300; % number of FGP iterations
+epsil_tol =  1.0e-05; % tolerance
+eta =  0.2; % Reference image gradient smoothing constant
+tic; u_fgp_dtv_g = FGP_dTV_GPU(single(vol3D), single(vol3D_ref), lambda_fgp, iter_fgp, epsil_tol, eta); toc; 
+figure; imshow(u_fgp_dtv_g(:,:,15), [0 1]); title('FGP-dTV denoised volume (GPU)');
+%%
\ No newline at end of file
diff --git a/Wrappers/Matlab/demos/demoMatlab_denoise.m b/Wrappers/Matlab/demos/demoMatlab_denoise.m
index 7f87fbb..145f2ff 100644
--- a/Wrappers/Matlab/demos/demoMatlab_denoise.m
+++ b/Wrappers/Matlab/demos/demoMatlab_denoise.m
@@ -1,5 +1,6 @@
 % Image (2D) denoising demo using CCPi-RGL
-
+clear
+close all
 addpath('../mex_compile/installed');
 addpath('../../../data/');
 
@@ -8,31 +9,55 @@ u0 = Im + .05*randn(size(Im)); u0(u0 < 0) = 0;
 figure; imshow(u0, [0 1]); title('Noisy image');
 
 %%
-fprintf('Denoise using ROF-TV model (CPU) \n');
+fprintf('Denoise using the ROF-TV model (CPU) \n');
 lambda_rof = 0.03; % regularisation parameter
 tau_rof = 0.0025; % time-marching constant 
 iter_rof = 2000; % number of ROF iterations
 tic; u_rof = ROF_TV(single(u0), lambda_rof, iter_rof, tau_rof); toc; 
 figure; imshow(u_rof, [0 1]); title('ROF-TV denoised image (CPU)');
 %%
-% fprintf('Denoise using ROF-TV model (GPU) \n');
+% fprintf('Denoise using the ROF-TV model (GPU) \n');
 % lambda_rof = 0.03; % regularisation parameter
 % tau_rof = 0.0025; % time-marching constant 
 % iter_rof = 2000; % number of ROF iterations
 % tic; u_rofG = ROF_TV_GPU(single(u0), lambda_rof, iter_rof, tau_rof); toc;
 % figure; imshow(u_rofG, [0 1]); title('ROF-TV denoised image (GPU)');
 %%
-fprintf('Denoise using FGP-TV model (CPU) \n');
+fprintf('Denoise using the FGP-TV model (CPU) \n');
 lambda_fgp = 0.03; % regularisation parameter
 iter_fgp = 1000; % number of FGP iterations
-epsil_tol =  1.0e-05; % tolerance
+epsil_tol =  1.0e-06; % tolerance
 tic; u_fgp = FGP_TV(single(u0), lambda_fgp, iter_fgp, epsil_tol); toc; 
 figure; imshow(u_fgp, [0 1]); title('FGP-TV denoised image (CPU)');
 %%
-% fprintf('Denoise using FGP-TV model (GPU) \n');
+% fprintf('Denoise using the FGP-TV model (GPU) \n');
 % lambda_fgp = 0.03; % regularisation parameter
 % iter_fgp = 1000; % number of FGP iterations
 % epsil_tol =  1.0e-05; % tolerance
 % tic; u_fgpG = FGP_TV_GPU(single(u0), lambda_fgp, iter_fgp, epsil_tol); toc; 
 % figure; imshow(u_fgpG, [0 1]); title('FGP-TV denoised image (GPU)');
 %%
+fprintf('Denoise using the FGP-dTV model (CPU) \n');
+% create another image (reference) with slightly less amount of noise
+u_ref = Im + .01*randn(size(Im)); u_ref(u_ref < 0) = 0;
+% u_ref = zeros(size(Im),'single'); % pass zero reference (dTV -> TV)
+
+lambda_fgp = 0.03; % regularisation parameter
+iter_fgp = 1000; % number of FGP iterations
+epsil_tol =  1.0e-06; % tolerance
+eta =  0.2; % Reference image gradient smoothing constant
+tic; u_fgp_dtv = FGP_dTV(single(u0), single(u_ref), lambda_fgp, iter_fgp, epsil_tol, eta); toc; 
+figure; imshow(u_fgp_dtv, [0 1]); title('FGP-dTV denoised image (CPU)');
+%%
+% fprintf('Denoise using the FGP-dTV model (GPU) \n');
+% % create another image (reference) with slightly less amount of noise
+% u_ref = Im + .01*randn(size(Im)); u_ref(u_ref < 0) = 0;
+% % u_ref = zeros(size(Im),'single'); % pass zero reference (dTV -> TV)
+% 
+% lambda_fgp = 0.03; % regularisation parameter
+% iter_fgp = 1000; % number of FGP iterations
+% epsil_tol =  1.0e-06; % tolerance
+% eta =  0.2; % Reference image gradient smoothing constant
+% tic; u_fgp_dtvG = FGP_dTV_GPU(single(u0), single(u_ref), lambda_fgp, iter_fgp, epsil_tol, eta); toc; 
+% figure; imshow(u_fgp_dtvG, [0 1]); title('FGP-dTV denoised image (GPU)');
+%%