多普勒盲区条件下的改进粒子滤波跟踪算法

机载预警雷达采用脉冲多普勒体制，具有良好的低空探测性能，但其存在不可忽略的多普勒盲区问题。在目标跟踪的过程中，该盲区容易造成目标航迹中断和重起批。针对多普勒盲区条件下的目标连续跟踪问题，提出了一种基于多假设运动模型的改进粒子滤波跟踪方法。该方法根据多普勒盲区对目标状态的约束形成多个可能的运动模型，粒子在每个模型下再进行带有状态约束的更新，当新出现的量测值落入任何一个运动模型形成的粒子云内，则航迹关联成功。仿真结果表明，该算法在不同盲区范围条件下，针对不同机动能力的目标均具有较高的航迹关联率，能够实现目标的连续跟踪。

An Improved Method for Tracking Targets in Doppler Blind Zone Based on Particle Filter

Airborne early warning radar with pulse Doppler (PD) system has good detection performance for low altitude targets, but the problem about Doppler blind zone (DBZ) can’t be ignored. In the process of tracking targets, the blind zone can cause interruption of target tracking termination and new initiation of track easily. Considering the problem of continuous target tracking in DBZ, an improved particle filter (PF) algorithm based on multiple hypothesis motion models (MHMM) is proposed. The algorithm produces multiple possible motion models according to target state constraint decided by Doppler blind zone. In each motion model, the particles are updated within state constraint. Once the first measurement after the reappearance of the target from DBZ falls into particle cloud formed by a certain model, the track association is successful. The simulation results show that under the condition of different Doppler blind zone ranges, high track association ratio can be got in this algorithm for targets with different maneuverability. Accordingly this algorithm can be implemented in continuous target tracking.