Forest fires spread rapidly with severe consequences, making timely and effective fire monitoring and personnel rescue crucial. Unmanned Aerial Vehicle (UAV) swarms play a significant role in this scenario. To address the challenge of swarm control, we propose a centralized task assignment method with feedback based on behavior tree and communication link. This approach achieves collaborative task scheduling while preserving individual autonomy. For search behavior, we introduce a heat map search strategy suitable for vertical camera shots, along with optimal parameter selection. Experimental results demonstrate that the proposed swarm control method and optimized search strategy effectively accomplish the task, significantly reducing personnel rescue time by nearly half.