Multi‐sensor fusion for multi‐target tracking using measurement division

The iterated‐corrector probability hypothesis density (IC‐PHD) filter propagates the posterior intensity of each sensor at one time step to improve tracking accuracy. However, targets cannot be estimated by the IC‐PHD filter, if the detection probability of the last update sensor is low. To deal with this problem, this study presents a new multi‐sensor multi‐target tracking method. Analysing the iterative process of this filter, it can be observed that the measurements obtained by sensors can be divided into several measurement subsets. Then, the similarity among the measurements is described by two parts, the similarity by combining the posterior intensity updated by the measurement, and the credibility of the posterior intensity. The similarity can be used to determine whether the measurements are from one target and whether the measurement selection meets the real situation. Based on the similarity among measurements, a two‐way selection approach is presented to find out measurement subsets corresponding to the true targets. In this approach, measurement and measurement subset are mutually selected. Two measurements are also mutually selected. Thus only the measurement subsets corresponding to true targets have large weights. The simulation results show that the proposed method can reduce the miss‐detection and false alarm effectively.