/SCSA-2D2D-MKL

Primary LanguageC++MIT LicenseMIT

SCSA2D2D mage denoising algorithm [optimization]

The project is an optimization of the Semi-Classical Signal Analysis (SCSA) method for image denoinsing. The SCSA algorithm is used for signal and image processing. it maily decomposes the input image into the squared eigenfunctions parametrized by h, fe, and gm using the SCSA operator

This project accelerates the SCSA operator eigenfunctions decomposition using MKL parallezation.

COMPILATION

To build this program's executable file, please follow these steps:

  • load intel compiler : make sure you have intel compiler (icc) and intell MKL library installed on your environment, you can simply load these by typing:
$  module load intel/15

in case a different intel compiler is loaded, please edit the make.inc file to reflect the used compiler.

  • build/compile the project:
$ make

DEMO

```
$ MKL_NUM_THREADS=6 ./SCSA_2D2D_MKL --data data/img32_lena32_noisy.dat  data/img32_lena32.dat  -N 256 -d 4 -h 0.4 -gm 6 -fe 2
```
  • to change input parameters, type for example: 
    $ run_command.sh

IMPORTANT:

Please make sure that :

  • the input image is saved in binary format .dat file
  • the h parameter is a positive float value
  • the gm and fe parameters are integer value
  • The d parameter is an integer value divisible by 2