Track-Before-Detect Techniques

This chapter presents a review of several recent algorithms for track‐before‐detect (TBD) and assesses their performance, and presents an outline of the historical context of TBD. It also motivates the study of TBD with an example where conventional tracking fails. The chapter reviews a number of TBD algorithms: Baum Welch algorithm; Viterbi algorithm; the Particle Filter; Maximum‐Likelihood Probabilistic Data Association (PDA) algorithm; and Histogram‐Probabilistic Multi‐Hypothesis Tracking (H‐PMHT). It presents an example application of TBD within a maritime surveillance context. H‐PMHT is a unique method for TBD. The chapter focuses on the approach of TBD as a tracking/estimation problem rather than the detection problem perspective. Historically many TBD algorithms have been batch methods, requiring the collection of a sequence of data before the TBD algorithm is applied. The chapter presents an example application of TBD where substantial improvement is obtained over conventional tracking.