森林防火无人机系统设计与林火识别算法研究

为克服传统森林火灾监测手段监测范围有限、无法适用恶劣环境以及费用高昂等问题,提出了基于六旋翼无人机的林火监测系统.首先,采用开源架构,设计并研制用于控制无人机的自动驾驶仪;其次,搭建用于森林火情监控的小型六旋翼无人机.在传回的火情视频的基础上,利用视觉显著性检测方法,按照林火区候选、特征融合、特征分类、阈值判断等流程判断是否有林火发生.如果有火情出现,算法能够自动提取林火的形状、面积、蔓延速度等信息,并结合光电烟雾传感器实现火情预警.该系统具有垂直起降、对气象的要求低,且使用和维护成本低等特点.

Design of the UAV system for forest fire prevention and the research of the fire recognition algorithm

First,the autopilot of the unmanned aerial vehicle(UAV),based on the open source architecture,is designed and developed.Secondly,the small hex-copter UAV is built for the forest fire monitoring.The significant detection method is adopted to deal with the returned fire video,and the process such as the fire candidate region selection,feature fusion,feature classification,threshold judgment is employed to judge the occurrence of forest fires.If a fire occurs,the algorithm can compute the shape,area and spread speed and other information of the fire automatically,and start warning with the help of the photoelectric smoke sensor.The system has the characteristics of vertical takeoff and landing,low demand for weather,low use and maintenance costs.