Date: September 10, 2004

Time: 3:00 pm

Location: GCATT Room 325

Speaker(s): Michael Farrell

Title: Robust Automatic Clustering of Hyperspectral Imagery Using Non-Gaussian Mixtures

Abstract:
This talk addresses the utility of robust automatic clustering
of hyperspectral image data. Such clustering is possible only when the
background in a scene is accurately modeled. Mixtures of non-Gaussian
densities have been discussed recently, and here we move further down this
path. We derive an automated t mixture model for the background in
hyperspectral images, using two techniques for estimating parameters based
on the Expectation-Maximization algorithm. Visual and statistical
evaluation of these techniques are made with AVIRIS data. Dealing with the
data's inhomogeneity by developing proper models of the background (i.e.
clutter) in a hyperspectral image is important in target detection
applications, especially for accurate performance prediction and detector
analysis.

Bio:
Michael Farrell is a PhD candidate in the School of ECE at Georgia Tech. His current (thesis) research deals with performance, bounds, and selected applications of adaptive detection & estimation in remote sensing. While his main focus is on hyperspectral imaging, he has also dealt with other electro-optical data from IR sensors. Outside of GT, Michael has spent a significant amount of time at MIT Lincoln Laboratory working on signal processing projects in radar tracking and SAR imagery. Prior to earning his MS at Georgia Tech and his BS at the University of
Iowa, Michael grew up on milk, cheese, sausage, and beer in the great state of Wisconsin.

Date:	September 10, 2004
Time:	3:00 pm
Location:	GCATT Room 325
Speaker(s):	Michael Farrell
Title:	Robust Automatic Clustering of Hyperspectral Imagery Using Non-Gaussian Mixtures
Abstract:	This talk addresses the utility of robust automatic clustering of hyperspectral image data. Such clustering is possible only when the background in a scene is accurately modeled. Mixtures of non-Gaussian densities have been discussed recently, and here we move further down this path. We derive an automated t mixture model for the background in hyperspectral images, using two techniques for estimating parameters based on the Expectation-Maximization algorithm. Visual and statistical evaluation of these techniques are made with AVIRIS data. Dealing with the data's inhomogeneity by developing proper models of the background (i.e. clutter) in a hyperspectral image is important in target detection applications, especially for accurate performance prediction and detector analysis.
Bio:	Michael Farrell is a PhD candidate in the School of ECE at Georgia Tech. His current (thesis) research deals with performance, bounds, and selected applications of adaptive detection & estimation in remote sensing. While his main focus is on hyperspectral imaging, he has also dealt with other electro-optical data from IR sensors. Outside of GT, Michael has spent a significant amount of time at MIT Lincoln Laboratory working on signal processing projects in radar tracking and SAR imagery. Prior to earning his MS at Georgia Tech and his BS at the University of Iowa, Michael grew up on milk, cheese, sausage, and beer in the great state of Wisconsin.