Denoising Hyperspectral Image With Non-i.i.d. Noise Structure
The DATA folder contains two HSI data sets: Washington DC Mall and RemoteImage, which can be used in simulation experiment.
The six Demo_*.m files show different noise case in Washington DC Mall. You can just change variable 'filename' to see the performance in other dataset. More information can be found in:
[1] Chen Y, Cao X, Zhao Q, et al. Denoising hyperspectral image with non-iid noise structure[J]. IEEE transactions on cybernetics, 2017, 48(3): 1054-1066.
Citation: @article{chen2017denoising, title={Denoising hyperspectral image with non-iid noise structure}, author={Chen, Yang and Cao, Xiangyong and Zhao, Qian and Meng, Deyu and Xu, Zongben}, journal={IEEE transactions on cybernetics}, volume={48}, number={3}, pages={1054--1066}, year={2017}, publisher={IEEE} }
Hyperparameters and initial model parameters are robust for the performance, and generally do not need to be further tuned.