Linux/Unix>> LTDL>> 返回
项目作者： wc253

项目描述 ：
A Low-rank Tensor Dictionary Learning Method for Hyperspectral Image Denoising.
高级语言： MATLAB
项目地址: git://github.com/wc253/LTDL.git
创建时间： 2020-12-24T12:59:40Z
项目社区：https://github.com/wc253/LTDL
开源协议：
下载


Code: A Low-rank Tensor Dictionary Learning Method for Hyperspectral Image Denoising

All matlab codes of the paper TSP2020 A Low-rank Tensor Dictionary Learning Method for Hyperspectral Image Denoising.

pre_watercolors_MSI

Datasets

CAVE from here

ICVL from here. We downsample the ICVL datasets by msi=msi(1:2:size(msi,1),1:2:size(msi,2), :).

Jasper Ridge from here

Folder structure

  1. Demo_DL_syn.m                    : Detect the object 'road' on denoised jasperRidge HSIs via different methods (Fig. 7, 8). Please run it where we provide the pre‐computing denoising results and you can get the results in Fig. 7 and Fig. 8.
  2. Demo_denoise_ge.m                : Denoise the  CAVE-'watercolors' HSI with generated noise. It needs to take a lot of time so you can test all methods on a cropped HSI. Change noise level by modifying variables "sigma_ratio" in your experiments.
  3. Demo_denoise_v2.m                : Denoise the test ICVL HSIs and the jasperRidge HSI. Set “exp=0” to compare model driven methods with deep learning method (Table IV) and set 'exp=1' to denoise for target detection. To run the deep learning method in this demo, you should first download and install 'MatConvNet'. Please see 'Readme.txt' in the path 'lib\compete_methods\HSI‐DeNet1'.
  4. Demo_target_detection.m          : Evaluate the dictionary learning performance of LTDL with synthetic data (Fig. 4). You can see the pre‐computed results in the road of 'result\pre_synthetic_data_test_once'.
  5. data\
  6. ├────HSIDnet_data.mat              :the test ICVL HSI of HSI-DeNet
  7. ├────jasperRidge_10band.mat        :the jasperRidge2 HSI for detection
  8. ├────watercolors_MSI.mat           :a CAVE HSI
  9. lib\                              
  10. ├───LTDL_utilize\           : functions of the proposed LTDL method
  11. ├───hyperspectralToolbox\   : HSI detection toolbox https://github.com/isaacgerg/matlabHyperspectralToolbox
  12. ├───tensor_toolbox\         : tensor processing toolbox http://www.sandia.gov/~tgkolda/TensorToolbox/index‐2.5.html
  13. ├───tensorlab\              : tensor processing toolbox https://www.tensorlab.net/versions.html#3.0
  14. ├───quality_assess\         : functions of quality assessment indices http://gr.xjtu.edu.cn/web/dymeng
  15. ├───compete_methods\
  16. ├───────────────────ksvdbox\     : http://www.cs.technion.ac.il/~ronrubin/software.html
  17. ├───────────────────naonlm3d\    : http://personales.upv.es/jmanjon/denoising/arnlm.html
  18. ├───────────────────BM3D\        : http://www.cs.tut.fi/~foi/GCF‐BM3D/
  19. ├───────────────────BM4D\        : http://www.cs.tut.fi/~foi/GCF‐BM3D/
  20. ├───────────────────tensor_dl\   : http://gr.xjtu.edu.cn/web/dymeng
  21. ├───────────────────KBRreg\      : http://gr.xjtu.edu.cn/web/dymeng
  22. ├───────────────────LLRT\        : http://www.escience.cn/people/changyi/codes.html
  23. ├───────────────────HSI-DeNet1\  : http://www.escience.cn/people/changyi/codes.html
  24. ├───────────────────MStSVD\      : https://github.com/ZhaomingKong/Hyperspectral_Image_denoising
  25. ├───LRTA.m                       : http://gr.xjtu.edu.cn/web/dymeng
  26. ├───PARAFAC.m                    : http://gr.xjtu.edu.cn/web/dymeng
  27. ├───myPlotROC.m                  : plot ROC curves
  28. ├───tight_subplot.m              : create "subplot" axes with adjustable gaps and margins
  29. result\ 
  30. ├──────pre_jasperRidge_10band    : the pre-computing results of 'Demo_denoise_v2' for MSI detection
  31. ├──────pre_synthetic_data_test_once
  32. ├──────pre_watercolors_MSI

Citation

X. Gong, W. Chen and J. Chen, “A Low-Rank Tensor Dictionary Learning Method for Hyperspectral Image Denoising,” in IEEE Transactions on Signal Processing, vol. 68, pp. 1168-1180, 2020, doi: 10.1109/TSP.2020.2971441.

We would like to thank those researchers for making their codes and datasets publicly available. If you have any question, please feel free to contact me via: xiaogong@bjtu.edu.cn