A joint tensor-based model for hyperspectral anomaly detection
2021
Zhang, Lili | Cheng, Baozhi
Nowadays, some algorithms based on tensor have drawn increasing attention in hyperspectral image (HSI) analysis. In this paper, a joint tensor-based model (JTen) is proposed for hyperspectral anomaly detection. First, the norm-2 between the original dataset and its projection onto the background dictionary tensor subspace is calculated. Then, low-rank and sparse matrix decomposition is employed and the sparse matrix that mainly contains the information of anomaly targets is got. The sparse matrix is treated as a new test dataset, and the norm-2 between the sparse matrix and the projection onto its background dictionary tensor subspace is calculated. Finally, the two norm-2 are joined by a coefficient and the detection result is obtained by using the joint tensor-based model. Experiments are carried out on real and synthetic HSI, and the results show that the proposed JTen generally outperforms the comparison algorithms.
显示更多 [+] 显示较少 [-]