Pytorch implementation of TIP paper "Binary Change Guided Hyperspectral Multiclass Change Detection"
Binary Change Guided Hyperspectral Multiclass Change Detection
Please cite our paper if you find it useful for your research.
@ARTICLE{10011164, author={Hu, Meiqi and Wu, Chen and Du, Bo and Zhang, Liangpei}, journal={IEEE Transactions on Image Processing}, title={Binary Change Guided Hyperspectral Multiclass Change Detection}, year={2023}, volume={32}, number={}, pages={791-806}, doi={10.1109/TIP.2022.3233187}}
Install Pytorch 1.10.2 with Python 3.6
Download the data file, including input data and training samples,passcode提取密码：8z8a, Baidu Netdisk, Link：https://pan.baidu.com/s/1AVG7YhU1e9NYSgcruL7PcQ code：h3ul
or google drive,Link: https://drive.google.com/drive/folders/1qxtbLm4zu6pNvN25ypfarssGSSJV7U-B.
ChinaData
China_MultChange.mat (input data)
X_3d：channel, height, width
Y_3d：channel, height, width
endmember：channel, num_em
Two_CMap：height, width （0 means unchanged, 1 means changed） Mul_CMap：height, width （0 means unchanged, 1,2,3... means different changed class）
China_sampIdx_16384.mat (used for pre-training of United Unmixing Module.)
idx_sample： 1, 16384 （used for python index； selected for pre-training of United Unmixing Module）
China_sampIdx_12288_4096.mat (for pre-training of Temporal Correlation Module and alternative optimization of the two modules. )
binary_label, 1,12288（the first 8192 samples are unchanged（labeled value as 0）; the remained 4096 samples are changed, labeled value as 1）
idx_sample: 1, 12288（index of the samples, used for python index； selected for pre-training of Temporal Correlation Module and alternative optimization of the two modules ）
maincode.py
🌷Homepage🌷
🔴Google web 🔴
🌏ResearchGate 🌍