<p>Coverage path planning (CPP) for unmanned aerial vehicles (UAVs) plays a crucial role in intelligent distributed surveillance systems, with applications in environmental monitoring, disaster response, and infrastructure inspection. However, many existing CPP approaches often lead to significant trajectory overlaps, unvisited areas, or even collisions in complex environments, resulting in prolonged task completion times and reduced coverage efficiency. This paper addresses the long-term CPP problem, where multiple UAVs collaborate to fully cover a known complex environment with a stochastic number of UAVs, aiming to balance the workload among them and promote fairness. We propose a novel framework called STC-Transfer, to tackle this problem. Initially, the workspace is partitioned among UAVs using a breadth-first search approach based on environmental data. Then, the transfer algorithm redistributes the coverage areas among UAVs to balance the workload and prevent over-concentration in specific regions. A coverage path is constructed within each assigned sub-area, where optimal trajectories are computed to minimize travel distance while ensuring complete coverage, based on an enhanced version of the spanning tree coverage algorithm. Finally, the weighted transfer algorithm enables immediate reassignment of coverage areas when the number of active UAVs changes, ensuring seamless task execution even when UAVs join or leave the mission unexpectedly. Experiments are conducted on various environments based on real-world city maps, where the number of UAVs varies over time. The results demonstrate the superior efficiency of the proposed framework compared to benchmark algorithms in terms of coverage area assignment, coverage path length, and runtime.</p>