Realistic extended target model for track before detect in maritime surveillance

Traditional target tracking algorithms are generally fed a set of thresholded detections under the hypothesis that no more than one detection is generated by each single target. Improvements in modern radar systems have made possible to obtain high resolution data of the targets, making them occupy more than a single resolution cell, and have made necessary to use appropriate Extended Target Tracking (ETT) techniques, since the aforementioned hypothesis is no longer valid. However, these techniques do not often take into account the physical phenomena that are involved in the radar-sea-target system. This paper explores some of the effects involved in the transmission, propagation, backscattering and processing of the radar signal in a maritime environment, that have a crucial importance in ETT. A statistical model that considers the featured effects is developed and tested in a particle filter based Track before Detect (TbD) algorithm. Accounting for physical aspects, good outcomes in both kinematic (i.e. position and velocity) and size (i.e. width and length) estimation can be pointed out using real radar data acquired by a high resolution X-band Marine radar located in the Gulf of La Spezia, Italy.