Hyperspectral anomaly detection by isolation using nearest neighbor ensembles

Xiangyu Song

doi:10.1117/12.3026716

21 February 2024 Hyperspectral anomaly detection by isolation using nearest neighbor ensembles

Xiangyu Song

Proceedings Volume 13080, International Conference on Frontiers of Applied Optics and Computer Engineering (AOCE 2024); 1308009 (2024) https://doi.org/10.1117/12.3026716
Event: International Conference on Frontiers of Applied Optics and Computer Engineering, 2024, Kunming, China

Abstract

In this paper, we proposed a novel isolation-based hyperspectral anomaly detector using nearest neighbor ensemble (iNNE) based on the premise that anomaly pixels are more prone to isolation compared to the background. The approach serves as an effective anomaly detector relying on nearest neighbors and isolation. iNNE demonstrates a notable enhancement in computational efficiency compared to established nearest neighbor-based methods, such as the Local Outlier Factor (LOF), especially when applied to datasets with thousands of dimensions or millions of instances. This improved efficiency is attributed to the method's linear time complexity and constant space complexity. In contrast to existing tree-based isolation methods like iForest algorithm, the proposed approach effectively addresses the challenges of detecting local anomaly instances and anomalies in high-dimensional data. The experimental outcomes on four authentic hyperspectral datasets have illustrated that the proposed iNNE not only outperforms other state-of-the-art anomaly detectors in terms of performance but also proves to be quite competitive in terms of computational efficiency.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Xiangyu Song "Hyperspectral anomaly detection by isolation using nearest neighbor ensembles", Proc. SPIE 13080, International Conference on Frontiers of Applied Optics and Computer Engineering (AOCE 2024), 1308009 (21 February 2024); https://doi.org/10.1117/12.3026716

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available