ftmlik/acda

Pytorch code of "Hyperspectral Anomaly Change Detection Based on Auto-encoder"

ACDA

This is a code of the paper "Hyperspectral anomaly change detection based on autoencoder" implemented on PyTorch. Pytorch is needed for running this code.

Paper is available on https://ieeexplore.ieee.org/abstract/document/9380336

My personal google web：https://scholar.google.com.hk/citations?hl=zh-CN&user=jxyAHdkAAAAJ

-=-=-=-=-=-=-=-=-=-=-=-=-=-==-=-=-=-=-==-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- [Dataset]: "Viareggio 2013" with de-striping, noise-whitening and spectrally binning

1. img_data.mat:
   img_1(D1F12H1); img_2(D1F12H2); img_3(D2F22H2)

链接：https://pan.baidu.com/s/1sRmdjsT-xl6DQJeoPIBNYA 提取码：qdqf

2. pretrain_samples:
   un_idx_train1,un_idx_valid1,un_idx_train2,un_idx_valid2; [acquired from the pre-detection result of USFA, Wu C, Zhang L, Du B. Hyperspectral anomaly change detection with slow feature analysis[J]. Neurocomputing, 2015, 151: 175-187.]
3. groundtruth_samples:
   un_idx_train1,un_idx_valid1,un_idx_train2,un_idx_valid2;
4. random_samples: un_idx_train1,un_idx_valid1,un_idx_train2,un_idx_valid2;

-=-=-=-=-=-=-=-=-=-=-=-=-=-==-=-=-=-=-==-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- [Usage]: maincode.py

-=-=-=-=-=-=-=-=-=-=-=-=-=-==-=-=-=-=-==-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

If you use this code for your research, please cite our papers: Hu M, Wu C, Zhang L, et al. Hyperspectral anomaly change detection based on autoencoder[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14: 3750-3762.