Improved CPHD filtering with unknown clutter rate

To accommodate the model mismatch in clutter rate, a cardinality probability hypothesis density (CPHD) filter with unknown clutter rate has been proposed by Mahler. It has proved to be a promising algorithm for multi-target tracking in complex environment. However, in Mahler's algorithm, the calculation of the number of clutters without observations is determined by the hybrid cardinality distribution and hybrid probability of misses, it will cause the confusion between undetected targets and clutters. To solve this problem, an improved CPHD filter is proposed which increases an estimation of the number of targets based on the measurement likelihood in the process of update and then modifies the hybrid cardinality distribution by treating the confused targets as detected ones more reasonably. Simulation results show that the improved CPHD filter is superior to the traditional method in both the estimates of clutter number and target state.