Considering various satellite constraints, target constraints and data transmission constraints, the mission planning problem is abstracted into a combinatorial optimization problem. In order to meet the requirement of repeatability of planning results in engineering practice, an Earth observation task planning algorithm based on improved dynamic programming algorithm is designed. The algorithm uses window preprocessing to reduce the search space, the target circular arrangement strategy to improve the global resource allocation, and the local greedy strategy to ensure the optimal execution time. The experimental results show that the planning results are better than the greedy planning algorithm, and the planning algorithm is feasible and effective in solving the problem of multi-agile satellite multi-constraint Earth observation mission planning.