To meet the requirements of air–sea cooperative interception in maritime blockade missions, and to address the difficulty of conventional multi-unmanned aerial vehicle path planning in handling both complex path constraints and continuous energy limitations, a two-stage optimization-based method for path planning and charging scheduling is proposed. In the first stage, a behavior-based directed acyclic graph is constructed in the action–state space, and the joint path planning problem is formulated as an integer programming model with task time window and battery-energy constraints, thereby generating task routes and coarse sea-based charging decisions for each UAV. In the second stage, a mixed-integer linear programming model is built on this basis to refine the docking-and-charging duration within given charging windows, so as to reduce overall task execution time as much as possible while ensuring that the onboard batteries never become fully depleted. Numerical simulation results in the designed maritime confrontation scenarios with different scales of hostile unmanned surface vessels (USVs) indicate that the proposed method can effectively intercept most hostile USVs, demonstrating the effectiveness of the two-stage optimization framework for air–sea cooperative interception.