<p>Space-based multi-target tracking is one of the key enabling technologies for space situational awareness. As to solve the multi-target passive tracking problem by a cooperative dual-observer system, a centralized parallel fusion method that considers measurement association probability is developed. Firstly, an orbital dynamics model perturbed by Earth’s oblateness and lunar gravity, together with a line-of-sight (LOS) angles measurement model are established. Secondly, a new centralized parallel fusion multi-target tracking frame considering measurement association probabilities is designed for dual-observer cooperative system. In detail, a dual-observer LOSs association evaluation metric and a two-stage measurements-to-measurements association algorithm are developed, where the measurement association probabilities are calculated to achieve high-precision cooperative data association. Subsequently, a centralized parallel tracking filter is proposed for accurately estimating the targets’ orbital states based on the frame of probability hypothesis density (PHD) filter. Multiple sets of augmented measurements based on the measurement association results are constructed, of which the associated measurement probabilities are incorporated into the Gaussian Mixture PHD (GM-PHD) filter to adjust the weights of the Gaussian components. Finally, numerical simulations are conducted with a GEO-type mission, where a 10-target’s tracking case is taken for validation. The results show that the accurate tracking is achieved, where the performance advantage is analyzed and confirmed by comparing with traditional distributed tracking methods.</p>