Ensemble and random collaborative representation-based anomaly detector for hyperspectral imagery

Yihang Lu, Xuan Zheng, Haonan Xin, Haoliang Tang, Rong Wang, Feiping Nie

Abstract: Highlights•Utilizing the random sub-sampling results as the background pixels instead of the sliding dual window trick.•Significantly reducing computational complexity.•Introducing iForest to HAD which provides complementary information to improve the accuracy and robustness.•Experiments illustrate the superiority and effectiveness of our proposed method.