摘要

This letter presents a theoretical comparative study of the performance bounds for point and extended target tracking. As a result of the additional information provided by the high-resolution sensors, the posterior Cramer-Rao lower bound (PCRLB) of the kinematic target states is proven to be always smaller under the extended target tracking framework than the point target model. Three assumptions about the two kinds of target models, which are satisfied in most target tracking problems, are presented as a sufficient condition of the inequality. This comparison result suggests the use of the extended target model to potentially achieve better performance in tracking applications. The superior performance bound is also illustrated with an elliptically modeled extended target tracking example.