csjunxu / guided-image-denoising-tip2018 Goto Github PK
View Code? Open in Web Editor NEWExternal Prior Guided Internal Prior Learning for Real-World Noisy Image Denoising. IEEE Transactions on Image Processing, 2018.
License: Other
External Prior Guided Internal Prior Learning for Real-World Noisy Image Denoising. IEEE Transactions on Image Processing, 2018.
License: Other
% =============================================================== The code in this package implements the Guided Image Denoising method for image denoising as described in the following paper: Jun Xu, Lei Zhang, and David Zhang External Prior Guided Internal Prior Learning for Real-World Noisy Image Denoising. IEEE Transactions on Image Processing (TIP), 2018. Please cite the paper if you are using this code in your research. Please see the file License.txt for the license governing this code. Version: 1.0 (03/28/2018), see ChangeLog.txt Contact: Jun Xu <[email protected]/[email protected]> % =============================================================== Note ------------ Please refer to https://github.com/csjunxu/GID_TIP2018 for the updates of the code. Overview ------------ Training Code: The code for learning external prior is provided in the folder "PG-GMM_TrainingCode", which relies on the training images in the subfolder "Kodak24" (please refer to the "Data" section). Testing Code: The function "Demo_Guided_DND2017" demonstrates real-world image denoising with the Guided Image Denoising method introduced in the paper. The function "Demo_Guided" demonstrates real-world image denoising with "ground truth" by the Guided Image Denoising method introduced in the paper. The function "Demo_Guided_NoGT" demonstrates real-world image denoising without "ground truth" by the Guided Image Denoising method introduced in the paper. Model ------------ The trained model "PGGMM_RGB_6x6_3_win15_nlsp10_delta0.001_cls33.mat" can be downloaded from https://github.com/csjunxu/GID_TIP2018/PG-GMM_TrainingCode Data ------------ Please download the data from corresponding addresses. 1. Kodak24: 24 high quality color images from Kodak PhotoCD dataset This dataset can be found at http://r0k.us/graphics/kodak/ 2. NCImages: real-world noisy images with no ''ground truth'' from "NoiseClinic" This dataset can be found at http://demo.ipol.im/demo/125/ The "CCImages" directory include two parts: 3. CC15: 15 cropped real-world noisy images from CC [1]. This dataset can be found at http://snam.ml/research/ccnoise The smaller 15 cropped images can be found on in the directory ''Real_ccnoise_denoised_part'' of https://github.com/csjunxu/MCWNNM_ICCV2017 The *real.png are noisy images; The *mean.png are "ground truth" images; The *ours.png are images denoised by CC. 4. CC60: 60 cropped (by us) real-world noisy images from CC [1]. "CC_60MeanImage" inlcudes the "ground truth" images; "CC_60NoisyImage" inlcudes the noisy images; 5. DND_2017: 1000 cropped real-world noisy images from DND [2]. Please download the dataset from https://noise.visinf.tu-darmstadt.de/ and put the files in "DND_2017" directory accordingly. 6. PolyUImages: 100 cropped images from our new dataset. The *real.JPG are noisy images; The *mean.JPG are "ground truth" images; [1] Seonghyeon Nam*, Youngbae Hwang*, Yasuyuki Matsushita, Seon Joo Kim. A Holistic Approach to Cross-Channel Image Noise Modeling and its Application to Image Denoising. CVPR, 2016. [2] Tobias Pl?tz and Stefan Roth. Benchmarking Denoising Algorithms with Real Photographs. CVPR, 2017. Dependency ------------ This code is implemented purely in Matlab2014b and doesn't depends on any other toolbox. Contact ------------ If you have questions, problems with the code, or find a bug, please let us know. Contact Jun Xu at [email protected] or [email protected]
Hi Jun,
Thank you for sharing the code and data sets!
Can you please explain on a high level what is the difference between 'Denoising_Guided_EI' and 'Denoising_Guided_NoGT' -- is it just you are providing the sigma i.e. the noise level in the first one? I also see '% PGPD denoising' and '% Guided denoising' in the comments, can you please explain what is PGPD -- by guided I understand external prior guiding the internal prior.
Also, in one of the demos 'NoiseEstimation' is called but in others it is not.
Looking forward to hear from you.
Thanks,
Touqeer
"Y_hat(:,blk_arr(:,idx)) = Y_hat(:,blk_arr(:,idx)) + bsxfun(@plus, Dnew*alphanew, DC(:,idx));"
这个语句,当blk_arr(:,idx)的值相同时,Y_hat 会取出同一列,但是求和操作只起一次作用。
举例:a=zeros(3,3);b=ones(3,2);index=[1,1];
a(:,index)=a(:,index)+b;
a =
1 0 0
1 0 0
1 0 0
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.