A hierarchical trajectory planning method based on improved particle swarm optimization(PSO) combined with Gauss pseudo-spectral method is proposed to solve the trajectory planning of multi-target reconnaissance missions for unmanned aerial vehicles (UAVs). Firstly, an improved PSO algorithm is designed for trajectory pre-planning. Aiming at the problem of slow convergence of traditional PSO algorithms, which is easy to fall into the local optimization problem, the chaos mapping initialization and adaptive parameter adjustment strateges are introduced to accelerate the algorithm convergence and improve the optimality of the solution. On this basis, combined with the shortest path solution strategy, the sequence of reconnaissance tasks is completed, and the executable reconnaissance task plan is obtained. Furthermore, in order to improve the accuracy and controllability of reconnaissance trajectories, trajectory re-planning based on Gauss pseudo-spectral method is proposed. By introducing the results of trajectory pre-planning as initial value guessing, the efficiency and accuracy of the algorithm are further improved. Finally, the simulation under complex environments shows that the designed algorithm can quickly generate mission execution plans and high-precision reconnaissance trajectories, which is consistent with the engineering applications of military UAVs performing reconnaissance tasks.