Magnutic / HazeRemoval

Single-Image Haze Removal Implementation

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Haze Removal

This is an Implementation of Fast Single-Image Haze Removal Algorithm Using Color Attenuation Prior [1], done as a term project for an image processing class. It also includes an implementation of Guided Image Filtering [2].

Implemented in C++14, using DevIL for loading and saving images.

License

ZLib License, see LICENSE.txt

Building

CMake is used to generate builds.

Linux, Makefiles

Install DevIL from your package manager, e.g. for Debian-based distros:

$ sudo apt-get install libdevil-dev

Then create a build directory, navigate to it, and invoke CMake:

$ cmake <haze-removal-root-dir>

Build using make:

$ make -j4

Windows, Visual Studio

CMake's find_package(DevIL) seems to have some problems on Windows. A working approach is to manually specify include directory and library files:

Download the DevIL Windows SDK from e.g. the DevIL sourceforge page. Extract to some directory, e.g. C:\lib\. Create a build directory, navigate to it, then invoke CMake:

$ cmake <haze-removal-root-dir> -DIL_INCLUDE_DIR="C:\lib\DevIL Windows SDK\include" -DIL_LIBRARIES="C:\lib\DevIL Windows SDK\lib\x86\Release\DevIL.lib" -DILU_LIBRARIES="C:\lib\DevIL Windows SDK\lib\x86\Release\ILU.lib"

Replace C:\lib\ with wherever you extracted the DevIL SDK, replace x86 with x64 if you are doing a 64-bit build. CMake will then produce a Visual Studio solution.

The DLL files provided with the SDK need to be either copied into the same directory as the resulting .exe file, or into a directory in the PATH environment variable.

References

[1] Q. Zhu, J. Mai and L. Shao, "A Fast Single Image Haze Removal Algorithm Using Color Attenuation Prior," in IEEE Transactions on Image Processing, vol. 24, no. 11, pp. 3522-3533, Nov. 2015. doi: 10.1109/TIP.2015.2446191 URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7128396&isnumber=7131605

[2] K. He, J. Sun, X. Tang, "Guided Image Filtering", in European Conference on Computer Vision, 2010. URL: http://kaiminghe.com/publications/eccv10guidedfilter.pdf

About

Single-Image Haze Removal Implementation

License:zlib License


Languages

Language:C++ 95.1%Language:CMake 4.9%