Multiple-model multiple hypothesis probability hypothesis density filter with blind zone

According to the problem of continuous tracking of multiple manoeuvring targets under the blind zone, a multiple-model probability hypothesis density (MM-PHD) filter based on the multiple hypothesis method is proposed. Under the blind zone, target state is estimated using the multiple model method. The hypothesis state is as the update state in the update step. Once the measurement is received, it will be updated using the MM filter. An improved pruning and merging algorithm is proposed to solve the problem of the number of hypothesis Gaussian components increasing continuously. Simulation results show that the novel algorithm is more effective in tracking multiple manoeuvring targets and improves performance of continuous target tracking.