tehret / blind-denoising Goto Github PK
View Code? Open in Web Editor NEWModel-Blind Video Denoising Via Frame-to-frame Training
License: GNU Affero General Public License v3.0
Model-Blind Video Denoising Via Frame-to-frame Training
License: GNU Affero General Public License v3.0
Hey, I have read your paper and find that this study is very interesting and helpful.
When I try to run this code to see the blind denoising result, some wried thing happened.
First I download the video frame used in your paper according your guide. Here, I only download the station_2_1080p25.y4m and extract the frame to station_gray folder.
Then I get flow using TV-L1 Optical Flow Estimation method provided in your project.
Finally, I run the blinding_denoising.py script and all work.
The command line I use is :
cd tvl1flow
sh tvl1flow.sh station_gray/%03d.png 1 100 station_clean_flow/tvl1_%03d.flo
cd ..
python blind_denoising.py --input station_gray/%03d.png --ref station_gray/%03d.png --output res/%03d.png --flow station_clean_flow/tvl1_%03d.flo --last 100
The result seems fine. The PSNR l got is about 38db and the denoised image is fine.
But when i try to add some guassian noise with sigma 25 to each gray images, Things become different.
I didn't find the code for adding noise the video frame, so l write one myself and get noisy image in
station_noise folder.
The command line I use is :
cd tvl1flow
sh tvl1flow.sh station_noise/%03d.png 1 100 station_noise_flow/tvl1_%03d.flo
cd ..
python blind_denoising.py --input station_noise/%03d.png --ref station_gray/%03d.png --output noise25_res/%03d.png --flow station_noise_flow/tvl1_%03d.flo --last 100
I find that the wrapped image is all zero after wrapping the target image based on flow.
Naturally, the calculation of loss is no meaning and the final denosing result is a black image.
Part of images I use is posted below:
clean station image, 0001 frame:
noisy station image, 0001 frame, add guassian noise with sigma=25:
denoised image by your approach:
Can you help me point out the problem and give me some suggestion to get idea result in your paper.
Looking forward to your reply. ^_^
ubuntu 16.04
cmake 3.5.1
sudo apt-get install libpng-dev
sudo apt-get install libjpeg-dev
sudo apt-get install libtiff5-dev
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.