Theory of Particle Count Detection with an Application to the Triggering of Biological Warfare Detection Systems

摘要

A new procedure is presented for the detection of a bio-target signal in aerosol particle number count data when no prior knowledge of the existence of such a signal or of its characteristics (e.g., amplitude and shape) is assumed. Unlike previous bio-target detection algorithms, the algorithm in this paper is derived rigorously by the direct application of probability theory. To address the detection problem, probability theory is used to compare two models (or hypotheses); namely, a model (M1) that postulates the presence of a bio-target signal in the background interference, and an alternative model (M2) that postulates the presence of a bio-target signal in the background interference. The posterior probability for each model is calculated based on all the available prior information, and used to determine the posterior odds ratio 012 in favor of model M2 over model M1. The ratio provides a quantitative measure of the evidence for the presence of a bio-target signal in the data. The new detection algorithm has been applied to both simulated and real particle count data and found to perform well.