Level 2 Fusion: Situational Assessment Composition Fusion with Uncertain Classification
Stephen Craig Stubberud, Kathleen Ann Kramer · 2006
A Level 2 fusion approach to composition estimation with uncertain classification is described. The technique is based upon Markov chain estimation of composition that allows for measurements of class that are uncertain. For this approach, Bayesian taxonomy is used to represent the uncertainty. While previous work has address composition estimation under conditions where the detection of objects in a group is uncertain, composition when there is uncertainty of classification is a much more difficult problem to address. The Markov chain, while straightforward in its implementation, contains a significant complexity in its transition matrices that would be difficult to model in other standard approaches to estimation theory, such as the Kalman filter.