Epistemic decision theory applied to multiple-target tracking

Abstract

A decision philosophy that seeks the avoidance of error by trading off belief of truth and value of information is applied to the problem of recognizing tracks from multiple targets (MTT). A successful MTT methodology should be robust in that its performance degrades gracefully as the conditions of the collection become less favorable to optimal operation. By stressing the avoidance, rather than the explicit minimization, of error, we obtain a decision rule for trajectory-data association that does not require the resolution of all conflicting hypotheses when the database does not contain sufficient information to do so reliably. This rule, coupled with a set-valued Kalman filter for trajectory estimation, results in a methodology that does not attempt to extract more information from the database than it contains.

Procedurally rational decision-making and control

“…Let e denote the set of possible states of nature; e will often be a subset of m n October 1996 (n-dimensional Euclidean space). In our development it will be important to distinguish between the decision space and the state of nature; this will lead to an extension, due to Kenney [12,13], of the original epistemic utility theory developed in [6] and first applied in the engineering context in [14][15][16][17][18][19].…”

“…We present only the solution for d = 1 (the case for d > 1 may be found in [13]; we therefore associate kfwith k + 1. Under this scenario, (17) and (18) become…”

Extracting the fundamental frequency of a nonlinear chirp signal with modulated harmonic structure using ML, target tracking, and the Viterbi algorithm

“…The basic operation of the tracking draws from [6], and proceeds as follows. At each time epoch, a single detected frequencyf k is available as an input.…”

“…These spurious frequency detections are then filtered using one of two techniques, both of which exploit the fact that the fundamental frequency is a continuous function of time. The first technique is based on forming the sequence of detected frequencies into tracks, as in [6]. Spurious frequencies do not form into sustained tracks, and so can be removed.…”

Integrating Graph Partitioning and Matching for Trajectory Analysis in Video Surveillance

“…[44], [17], [24], [20], [34]. Recently, a class of techniques called "tracking by detection" has been shown to provide promising results [2], [27], [31], [9], [15]. For multi-object tracking (i.e.…”