Abstract | ||
---|---|---|
In this work we propose a novel unsupervised algorithm for designing multispectral filters that are tuned for local anomaly detection algorithms. This problem is formulated as a problem of channel reduction in hyperspectral images, which is performed by replacing subsets of adjacent spectral bands by their means. An optimal partition of hyperspec- tral bands is obtained by minimizing the Maximum of Maha- lanobis Norms (MXMN) of errors, obtained due to misrep- resentation of hyperspectral bands by constants. By mini- mizing the MXMN of errors, one reduces the anomaly con- tribution to the errors, which allows to retain more anomaly- related information in the reduced channels. We demonstrate that the proposed algorithm produces better results, in terms of the Receiver Operation Characteristic (ROC) curve of a benchmark anomaly detection algorithm (RX) - applied after the dimensionality reduction, as compared to two other di- mensionality reduction techniques, including Principal Com- ponent Analysis (PCA). |
Year | DOI | Venue |
---|---|---|
2009 | 10.5281/zenodo.41582 | EUSIPCO |
Keywords | Field | DocType |
hyperspectral imaging,object detection,optical filters,sensitivity analysis,signal detection,unsupervised learning,mxmn,pca,roc curve,adjacent spectral bands,anomaly contribution,anomaly-related information,benchmark anomaly detection algorithm,channel reduction,dimensionality reduction,hyperspectral bands,hyperspectral images,local anomaly detection algorithms,maximum of mahalanobis norms,multispectral filters,novel unsupervised algorithm,principal component analysis,receiver operation characteristic curve | Anomaly detection,Dimensionality reduction,Pattern recognition,Multispectral image,Communication channel,Mahalanobis distance,Hyperspectral imaging,Artificial intelligence,Spectral bands,Principal component analysis,Mathematics | Conference |
ISBN | Citations | PageRank |
978-161-7388-76-7 | 0 | 0.34 |
References | Authors | |
12 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Oleg Kuybeda | 1 | 3 | 1.20 |
David Malah | 2 | 219 | 60.95 |
Meir Barzohar | 3 | 94 | 11.06 |