Adaptive detection of low-observable targets in correlated sea clutter using Bayesian track-before-detect

Detection of small radar cross section targets embedded in sea clutter is a particularly demanding problem. One strategy is to lower the detection threshold which in turn increases the false alarm rate and stresses the tracking algorithm. Track-before-detect provides a way of enhancing the signal to interference ratio prior to thresholding, thereby improving the final detection performance. The adoption of a Bayesian filter for track-before-detect provides a natural extension to conventional statistical hypothesis-based detection. The method proposed in this paper avoids the `curse of dimensionality' of multi-target tracking by utilising an efficient localised target density formulation, that allows for simultaneous estimation of target and clutter parameters. This enables the algorithm to adapt the target and clutter models inherent in the likelihood functions to the actual data and improve the final detection performance. The goal of this work is to incorporate the spatial and temporal clutter correlation and target attributes into the Bayesian framework.