Aiming at several problems occurred in multi-target tracking, such as the moving targets interleaving or overlapping, and the target losing momentarily, an improved discrete continuous optimization for multi-target tracking algorithm is proposed. The method according to the principle of discrete continuous optimization for multiple target tracking, added speed constraints and ameliorated the original energy function, so as to achieve the purpose of constraint the trajectories form. Clustering strategies after global optimization is used to achieve the purpose of distinguishing different trajectories. The experimental results show that the algorithm has better robustness and can well realize multi-target tracking in complex image sequences.