Tracking of dense group targets based on motion grouping

To cope with the problem of numerous accompanied interference in the field of target detection and tracking in space, a fast detection and tracking method for targets in space based on dense multi-target motion grouping was proposed. Firstly, within the range allowed by the sensor resolution, the sparse optical flow was adopted to extract the motion information of the individual in the group, and then the generating function regularization was used to integrate the similarity between the motion paths. With the idea of “collective merging”, collective motions were detected from dense random motion, so that the group targets can be divided into several sparse groups with similar motion patterns in space. Finally, a graph model based on the topological relationship among sparse groups was constructed to filter out potential targets for which the false alarm was suppressed by inter-frame correlation. Simulation and experiment results show that the proposed method has good robustness and real-time performance for different group targets distribution in space.