search

danfenghong / IEEE_TNNLS_EGU-Net

Danfeng Hong, Lianru Gao, Jing Yao, Naoto Yokoya, Jocelyn Chanussot, Uta Heiden, Bing Zhang. Endmember-Guided Unmixing Network (EGU-Net): A General Deep Learning Framework for Self-Supervised Hyperspectral Unmixing, IEEE TNNLS, 2021.
55 stars 13 forks source link

How to get Trlabel? #9

Closed jikerWRN closed 2 years ago

jikerWRN commented 2 years ago

I understand that it has to be 8000x5 but the code in Pseudo_endmembers_generation.m line # 23 Abund = sunsal(EM, X', 'lambda', 0, 'ADDONE', 'no', 'POSITIVITY', 'yes', ... 'AL_iters', 200, 'TOL', 1e-4, 'verbose','yes');

EM here is around (8000xChannel) and X' (NxChannel) here is the data. How using Sunsal true label is generated to be 8000x5. This part is very confusing. The correct way should have been

Abund = sunsal(True_EM, EM, 'lambda', 0, 'ADDONE', 'no', 'POSITIVITY', 'yes', ... 'AL_iters', 200, 'TOL', 1e-4, 'verbose','yes');

Would you please clarify for me.

Regards,

Boinao

_Originally posted by @Boinao in https://github.com/danfenghong/IEEE_TNNLS_EGU-Net/issues/4#issuecomment-1006626463_

jikerWRN commented 2 years ago

Hi, I have the same question, I want to know how do you get the 'True_EM'? Can you get promising performances? Looking forwards to your reply! @Boinao