In this paper, NL-PCA is used as a first denoising stage to estimate spatially variant noise and to guide patch selection on the second stage (PRI-NLM3D). So basically they work in tandem, with the resulting two-stage algorithm called PRI-NL-PCA.
Which kind of problems are you facing with the code?
This is my file listing:
demo.m
cPRI_NL_PCA.mexw64
ssim_index3d.m
debias.p
t1_icbm_normal_1mm_pn0_rf0.rawb
NL_PCA.p demo.m
NoiseMapping.p
cNL_PCA.mexw64 medfilt3.m
I put all of them in a separate directory. Open MATLAB, change the working directory and run "demo.m". This works fine.
If you want to run the routines on your own images, you should convert them to double precision first. Just type "my_image = double(my_image)" in the command line. Otherwise the mexw64 compiled code will crash. I found execution times to be rather long on my I5 workstation, so you should first try denoising small volumes.
Which kind of problems are you facing with the code?
This is my file listing:
demo.m
cPRI_NL_PCA.mexw64
ssim_index3d.m
debias.p
t1_icbm_normal_1mm_pn0_rf0.rawb
NL_PCA.p demo.m
NoiseMapping.p
cNL_PCA.mexw64 medfilt3.m
I put all of them in a separate directory. Open MATLAB, change the working directory and run "demo.m". This works fine.
If you want to run the routines on your own images, you should convert them to double precision first. Just type "my_image = double(my_image)" in the command line. Otherwise the mexw64 compiled code will crash. I found execution times to be rather long on my I5 workstation, so you should first try denoising small volumes.
Thanks Mauro. The author have protected the source code in NL_PCA.p , cPRI_NL_PCA.mexw64 and i want to understand the steps of source code. Is there any way to understand the code?