The theory of single target tracking in complex target environments is already very comprehensive. Target tracking in complex environments nowadays can cause crossover and branching. Therefore, the application of this algorithm in engineering is relatively limited. Currently, the optimal Bayesian algorithm and multi fake method are considered the best methods for data interconnection, and many advanced radar systems at home and abroad have adopted this algorithm.

Multi target tracking technology involves two aspects: target state estimation and trajectory management. On the one hand, responsible for providing motion trajectory prediction estimates for tracking targets; The latter is responsible for the relationship between radar echoes and trajectory pairing, as well as the establishment and cancellation of trajectories. For filter design, it is not important to consider whether the filter can adapt to uniformly moving targets, but also to targets that are variable in speed or maneuvering.

Due to the correlation between measurement values and trajectories in trajectory correlation, it is necessary to know the prediction of the target’s arrival position at the current moment in the previous moment. If the prediction is incorrect, it is almost impossible to correctly correlate the radar measurement values and trajectories at the next moment. Therefore, target state estimation is the core issue in multi-target tracking technology, and establishing a correct target model is the primary consideration.

At present, the main negative impacts on tracking include clutter and false alarms, which cannot be effectively removed. Noise and false alarms can interfere with the accuracy of tracking. Therefore, when the source of observation cannot be determined, it is necessary to establish trajectory correlation between this observation and other data before conducting target state estimation.

The purpose of target state estimation is to establish or correlate the echoes of multiple targets detected by the radar in the current cycle with the trajectories of multiple targets, and filter out noise. Based on historical data and the updated trajectory data, it predicts the location of the next radar echo reception.