This article proposes a joint estimate approach in terms of target states and target cardinality as well as detection probability based on the Poisson multi-Bernoulli (PMB) filter. For the situations, where the detection probability is unknown or unreliable, the usual tracking methods often fail to work, so online estimation of the detection probability is indispensable.To depict the unknown detection probability, an inverse gamma Gaussian mixture (IGGM) implementation is adopted to propagate nonnegative features, including signal amplitude and signal-to-noise ratio (SNR). The IGGM-based PMB filter, abbreviated as IGGM-PMB, is proposed to solve the multitarget tracking (MTT) along with the unknown and time-varying detection profile. Specifically, the Poisson random finite set (RFS) intensity is approximated as an IGGM, while the density of Bernoulli RFS is approximated as a single inverse gamma Gaussian component (IGGC). Then, the proposed IGGM-PMB filter is applied to distributed multisensor fusion, wherein the estimated detection probabilities are used as a choice of the fusion ordering. Simulation results demonstrate the effectiveness and superiority of the proposed approach via comparisons with state-of-the-art approaches.

中文翻译：

---

基于IGGM的泊松多伯努利滤波器及其在分布式多传感器融合中的应用

本文提出了一种基于泊松多伯努利（PMB）滤波器的目标状态和目标基数以及检测概率的联合估计方法。对于检测概率未知或不可靠的情况，通常的跟踪方法往往不起作用，因此在线估计检测概率是必不可少的。为了描述未知的检测概率，采用逆伽马高斯混合（IGGM）实现传播非负特征，包括信号幅度和信噪比 (SNR)。提出了基于IGGM的PMB滤波器，缩写为IGGM-PMB，用于解决多目标跟踪（MTT）以及未知和时变检测轮廓。具体来说，泊松随机有限集 (RFS) 强度近似为 IGGM，而伯努利 RFS 的密度近似为单个逆伽马高斯分量 (IGGC)。然后，将所提出的 IGGM-PMB 滤波器应用于分布式多传感器融合，其中估计的检测概率用作融合排序的选择。仿真结果通过与最先进方法的比较证明了所提出方法的有效性和优越性。