hungpt297/dark-channel-prior-dehazing

Python implementation of the CVPR 2009 best paper "Single Image Haze Removal Using Dark Channel Prior"

##Dependencies

1. pillow(~2.6.0)
2. Numpy(~1.9.0)

If the scripts throw AttributeError: __float__, make sure your pillow has jpeg support e.g. try:

$ sudo apt-get install libjpeg-dev
$ sudo pip uninstall pillow
$ sudo pip install pillow

##How to generate the results

Enter the src directory, run python main.py. It will use images under img directory as default to produce the results. The results will show up in result directory.

To test special configurations for a given image, for example, to test the image with index 0 (check IMG_NAMES in util.py for indexes) and tmin = 0.2, Amax = 170, w = 15, r = 40, run

$ python main.py -i 0 -t 0.2 -A 170 -w 15 -r 40

Naming convetion of the results

For input image name.jpg using the default parameters, the naming convention is:

1. dark channel: name-dark.jpg
2. raw transmission map: name-rawt.jpg
3. refined tranmission map: name-refinedt.jpg
4. image dehazed with the raw transmission map: name-radiance-rawt.jpg
5. image dehazed with the refined transmission map: name-radiance-refinedt.jpg

If there are special configurations for the parameters, for example, , then the base name will be appended with -20-170-50-40 e.g. the dark channel is name-dark-20-170-50-40.jpg

##Directory structure

.
├─ README.md
├─ requirements.txt
├─ doc
│   └── report.pdf
├─ img (source images)
│   └── ... (input images from CVPR 09 supplementary materials)
├─ result (the results)
│   └── ...
└─ src (the python source code)
    ├── dehaze.py (dehazing using the dark channel prior)
    ├── main.py (generate the results for the report)
    ├── guidedfilter.py (guided filter)
    └── util.py (utilities)

##About

• Github repository
• Author: Qiuyi Zhang
• Time: Jan. 2015