Inspiration

Haze & Fog causes several Road and Plane accidents: Survey By Times of INDIA Shows about 11000 people died because of accidents met due to Fog/Haze during the year 2017 and the number keeps on increasing every year. Several Plane and Chopper crashes has been reported due to low visibility caused by Fog and Haze.In 2019 we lost Chief of Defence Staff,during a chopper crash caused by low visibility caused by haze. Another accident in which 234 people died during a indonesian Jet crash due to reduced visibility caused by Fog and Haze.

What it does

With the advancement in Image Processing Algorithms and Technology. We now have a solution the Dehaze_module implements the "Boundary Constraint and Contextual Regularization algorithm" which refers to theory and equations of the research by Gaofeng MENG. The module thus developed helps us to dehaze images. We can futher make use of the fact that video is made up of frames, we extract frames and process each frame and append it to a array and finally combine the processed frames to give Dehazed video output. It can be used to have a clear view in Haze/Foggy weather and reduce chances of road accidents and plane crashes. This project aims to Removes Haze and Fog and enhances quality of Videos/Images which in real time processes the video and gives clear view and prevents road accidents and plane crashes.

How we built it

 parmishhImage Dehazing Algorithm  parmishhVideo Dehazing Implementation

The concepts of the following algorithms and technologies were used: python, opencv, image-processing, dark-channel-prio,r dehazing-algorithm, boundary-constraint, contextual-regularization etc..

Challenges we ran into

It was challenging to host the project because it uses heavy processing but finally we figured out using streamlit and Replit. The understanding of image processing algorithms and implementing them were really time consuming, we ran into several errors and limitations but all effort were worth it and we finally achieved it.

Accomplishments that we're proud of

The outcome of the project is really satisfactory yet we still have scope of improvements. We are proud to publish this project as a PyPi package which can be easily installed from anywhere and the project can be used. Just use:

pip install Dehaze

To install this Project!

What we learned

we learned a lot of image processing algorithms before applying the research by Gaofeng MENG. we were engaged with Python, opencv and many other tools and technologies.

What's next for Dehazer

Some of the challenges for Dehazer is realtime processing as we need a good amount of processing power to process the results fast. However the good things is that there already exists technology that is capable for that.

Outputs

 parmishh  parmishh
 parmishh  parmishh

Additional Bg removal feature

 parmishh  parmishh
 parmishh  parmishh

Acknowledgements

Thanks to Gaofeng MENG, Ying WANG, Jiangyong DUAN, Shiming XIANG, Chunhong PAN for their paper "Efficient Image Dehazing with Boundary Constraint and Contextual Regularization"(https://ieeexplore.ieee.org/document/6751186)

Built With

  • boundary-constrains
  • contextual-regularization
  • dark-channel-priori
  • dehazing-algorithm
  • image-processing
  • opencv
  • python
  • video-processing
Share this project:

Updates