Bernoulli Multi-Target Track-Before-Detect for Maritime Radar

Detection and tracking of small maritime targets using high resolution radar is a challenging problem due to the temporally and spatially varying nature of the sea surface and the spiky returns which can resemble targets. The traditional approach to radar data processing is to first detect targets using their amplitude information (by comparing the radar returns to a suitably chosen threshold) and then feed the extracted (point) measurements into a target tracking system. An alternative approach which can be exploited to improve tracking of weak targets is to process the raw radar data directly though use of a track before detect (TBD) algorithm. This paper presents the development of a multi-target TBD algorithm for maritime radar with targets that are both widely separated and also closely spaced. The algorithm builds upon the Bayes-optimal single-target TBD filter, referred to as the Bernoulli TBD.