Realistic ship model for extended target tracking algorithms

Recent developments in high resolution sensors have encouraged the use of Extended Target Tracking (ETT) algorithms specifically designed to deal with targets that generate more than one detection per frame. At the same time, the availability of more powerful computational resources enable the use of soft computing techniques that yield a target probability, instead of a hard decision. This paper proposes a realistic target model feasible for an Extended Target - Track before Detect framework. Physical phenomena related to the acquisition of high resolution X-band marine radar data are considered. Real radar data is used to assess the superior performance of the featured model with respect to previous approaches. Results show that the realistic model provides better estimations of the target velocity and size.