Waterfall Solutions : Publication Abstract

Sorry for the mess

If you can see this text, it is likely that your browser is not honouring the Cascading Style Sheet (CSS) information upon which this site depends. As a result, the website probably looks very ugly indeed!

It may be possible to improve things by turning off the styling information altogether. To try this approach, click the button below and we will attempt to set a cookie that disables the CSS styling.

If this does not work, try scrolling down this page to the 'Change Style' section. Choose 'No Style' from the list and click on the 'Change Style' button.

Title

Hyperspectral anomaly detection and un-mixing in fat-tailed clutter

Abstract:

Detection of anomalies in hyperspectral clutter is an important task in military surveillance.

Most algorithms for unsupervised anomaly detection make either explicit or implicit assumptions about hyperspectral clutter statistics: for instance that the abundance is either normally distributed or elliptically contoured.

In this paper we investigate the validity of such claims. We show that while non-elliptical contouring is not necessarily a barrier to anomaly detection, it may be possible to do better.

In this paper we develop a method for visualizing the degree of elliptical contouring of hyperspectral data.

Having observed that most data fails to be elliptically contoured, we develop a new method for anomaly detection that has good performance on non-EC data.

Download:

Adobe Acrobat document, 440K

If you have problems viewing the PDF document, you may need to update Acrobat Reader to a more recent version. Click on the image below to visit the Adobe download site.

Authors:

M. Bernhardt, Waterfall Solutions, UK;
J. P. Heather, Waterfall Solutions, UK;
O. Watkins, Waterfall Solutions, UK;

Conference:

Joint Annual Technical Conference 2006, Edinburgh, UK, 13 - 14 July