kindraywind/singleimagehazeremover

A Python2 implementation of single image haze removal

Single image haze removal

A Python2 implementation of single image haze removal

The propose of this repository is to implement the image haze removal base on the Zhiming Tan Et al. paper.

Usage

The code can be executed via terminal

python dehaze.py then input the PATH_TO_IMAGE imagename.png

---

Sample input

Sample output

---

How does it work.

This dehaze algorithm contains three steps,

1. Determine intensity of atmospheric light
2. Estimate transmission map
3. Clarify image

First, the intensity of atmospheric light A is estimated form hazed image I(x). Then, the transmission map t(x) is estimated using A and I(x). Finally, the image is clarified with the image defogging model.

Step#1 Estimate intensity of atmospheric light:

Find the top 0.1% brightest pixels in the dark channel then choose one with highest intensity as the representing of atmospheric light.

Step#2 Estimate transmission map:

First, find a dark channel based on a local area(coarsemap) Then, the transmission map t(x) is thereby obtained:

t(x) = 1 – defoggingParam * darkPixelFromCoarseMap / AtmosphericLightIntensity

The defoggingParam is a value between 0 to 1. The higher value the lesser amount of fog would be kept for the distant objects.

Step#3 Clarify image:

Finally, the image is clarified by: J(x)=(I(x)- A)/max(t(x), t0)+A

Where J(x) is output, I(x) is input, t(x) is transmission map, A is atmospheric light and t0 is set to a constant value to avoid dividing by zero.

singleimagehazeremover's Issues

Dehaze.py takes forever to run

When running python dehaze.py <file_path> <image_name> it's taking forever for the code to execute. I kept it running on the terminal for 15 minutes and it didn't produce an output. Any reason why this could be happening?

Poosible to adapth this algorithm to work on huge images?

I have some large gigapixel panoramas (+/- 390Gb 16bit tiff) which need haze removal. Doing haze removal prior to stiching does not work well so haze removal has to be done after stiching. I can split the finished image in smaller pieces. Is it possible to adapt your algorithm to process all images and then do haze removal?