基于多帧状态估计机制的GM-PHD滤波器

GM-PHD Filter Based on Multi-frame State Estimation Mechanism

为处理低检测概率情况下目标漏检的情况,引入一种新的多帧状态估计机制,提出了一种基于多帧状态估计机制的高斯混合概率假设密度滤波器。该机制依据不同时间步骤的目标权值来构建每个目标的历史权值矩阵和状态提取标识符。在目标跟踪过程中,当一些连续运动目标在某些时间步骤漏检时,通过多帧状态估计机制,充分依据关联目标的权值矩阵和状态提取标识符来对目标的当前状态进行估计。仿真实验表明,所提算法在保证跟踪有效性的同时,能够在低检测概率且杂波率相对较高的情况下显著提高目标的跟踪性能,具有较强的鲁棒性。

A new multi-frame state estimation mechanism is introduced to deal with the situation of target undetected under low detection probability. A Gaussian mixed probability hypothesis density filter based on muhi-frame state estimation mechanism is proposed. The mechanism builds the historical weight matrix and the state extraction identifier for each target based on the target weights of the different time steps. In the process of target tracking, when a continuous-moving target is missed at some time steps, the current state of the target is estimated based on the weight matrix of the association target and the state extraction identifier, through the multi-frame state estimation mechanism. The simulation results show that the algorithm proposed has strong robustness, and can improve the target tracking performance greatly under the situation of low detection probability and relatively high clutter rate while ensuring the tracking effectiveness.