A Fast Method for Hyperspectral Target Detection with Matrix Inverse Updating and Downdating
- 10.2991/msam-18.2018.64How to use a DOI?
- remote sensing; matrix inverse; fast least squares; subpixel target detection
The unsupervised nonnegative constrained least squares (UNCLS) method is a well-known method for hyperspectral subpixel target detection. Depending on the complexity and dimensionality of the hyperspectral data, it may be computationally expensive for large matrix inversion with the UNCLS method. This paper proposed a fast UNCLS algorithm to improve its computational performance. Matrix inverse updating and downdating theory is implemented to speed up the iterative least square estimation in the UNCLS method. Experiments on both simulated and real hyperspectral images show that the proposed fast UNCLS can efficiently detect subpixel targets with precision guaranteed. We have demonstrated that our proposed method improves performance of the UNCLS method, which achieves a >10x speedup for both the simulated and real scenes when large number of targets are detected.
- © 2018, the Authors. Published by Atlantis Press.
- Open Access
- This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).
Cite this article
TY - CONF AU - Xin Wu AU - Mizhen Wang AU - Chen Yang PY - 2018/07 DA - 2018/07 TI - A Fast Method for Hyperspectral Target Detection with Matrix Inverse Updating and Downdating BT - Proceedings of the 2018 3rd International Conference on Modelling, Simulation and Applied Mathematics (MSAM 2018) PB - Atlantis Press SP - 301 EP - 307 SN - 1951-6851 UR - https://doi.org/10.2991/msam-18.2018.64 DO - 10.2991/msam-18.2018.64 ID - Wu2018/07 ER -