Detecting Small Low Emission Sources on a Large Random Background
February 02, 2011
12:00 p.m.
Peter Kuchment
Abstract
The talk will present the results of joint work with M. Allmaras, D. Darrow, Y. Hristova, and G. Kanschat.
It can be considered as a continuation of the study presented a year ago under the title “Deterministic ways of detecting small low emission sources on a large random background.” It addresses the problem of detecting a geometrically small low emission object in the presence of an overwhelming random background. For instance, the target SNR (signal-to-noise ratio) is on the level of 0.01%. Since last year’s presentation, several questions unresolved at the tie have been answered. In particular, now most of the deterministic (analytic) approach formerly used is eliminated and replaced by simple statistical considerations, which has lead to improvements in detection and possibility of estimating the relevant confidence probabilities.