EECS CSPL SEMINAR SERIES
WINTER TERM 1997


Alfred O. Hero, III

Alfred O. Hero, III

Dept. of Electrical Engineering and Computer Science

University of Michigan

hero@umich.edu



Tuesday, February 25, 1997
4:30-5:30 pm
1003 EECS


Object Detection and Recognition in Noisy Images: Part II, Robust Detection Techniques

Abstract -
This is the second of a series of talks on robust target detection and recognition in statistical imaging applications. In this talk we review methodologies for robust detection under nuisance parameters, apply methods of exact robustness to target detection in images with unknown clutter, show that under some conditions these methods are equivalent to well known CFAR methods for the case of homogeneous clutter, and derive novel CFAR tests for structured inhomogeneous clutter backgrounds. The novel tests are shown to outperform other CFAR methods over important ranges of SNR. We conclude with examples for the case that the target signature is known up to a complex scale factor, rotation, or translation, and when the target signature lies in a known set of possible target signatures.



For your information the following title and abstract is for the last of three talks on this topic, which I will be scheduling soon.


Object detection and recognition in noisy images: Part III, robust detection techniques

This is the last of series of talks on robust target detection and recognition in statistical imaging applications. In this talk we review methodologies for robust detection under nuisance parameters (local UMP, min-max, MP invariant), apply these methods to target detection in images with unknown clutter, show that these methods are equivalent to well known CFAR methods for the case of homogeneous clutter, and derive novel CFAR tests for structured inhomogeneous clutter backgrounds. The novel tests are shown to outperform other CFAR methods over important ranges of SNR. We conclude with examples in IR/SAR imaging when the target signature is known up to a scale factor and when the target signature lies in a set of known target signatures.


Biosketch -
Alfred O. Hero III was born in Boston, MA. in 1955. He received the B.S. summa cum laude from Boston University (1980) and the Ph.D. from Princeton University (1984), both in Electrical Engineering. He held the G.V.N. Lothrop Fellowship in Engineering at Princeton University. He is presently Associate Professor of Electrical Engineering and Computer Science and Area Chair for Signal Processing at the University of Michigan, Ann Arbor. He has held positions of Visiting Scientist at M.I.T. Lincoln Laboratory, Lexington, MA (1987 - 1989), Visiting Professor at Ecole Nationale de Techniques Avanc es (ENSTA), Paris, France (1991), and William Clay Ford Fellow at Ford Motor Company (1993). His research interests are in the areas of detection and estimation theory applied to statistical signal and image processing. Alfred Hero is a member of Tau Beta Pi, the New York Academy of Sciences, the American Statistical Association, and Commission C of the International Union of Radio Science (URSI). He was awarded the EECS Research Excellence Award at UM in 1995. He is currently an Associate Editor for the IEEE Transactions on Information Theory. He Chairs the Statistical Signal and Array Processing committee of the IEEE Signal Processing Society and is Conference Treasurer for the IEEE Signal Processing Society. He was Chairman for Publicity for the 1986 IEEE International Symposium on Information Theory. He was General Chairman for the 1995 IEEE International Conference on Acoustics, Speech, and Signal Processing.



return to Previous CSPL Seminars homepage