UAV-Based Intelligent System for Australian Flood Reconnaissance and Surveillance

Due to climate change in the past two years, Australia has often suffered from natural disasters, such as floods. Since last year, some regions of Australia have experienced four major flood crises. Besides, unmanned aerial vehicles (UAVs) have merits in terms of economy, safety, and operability, and there is a strong demand for UAVs in many civil fields. Small UAVs can be widely used in disaster prevention and mitigation, search and rescue, and other fields. Therefore, in this paper, a UAV-based framework is proposed for the detection, localization, segmentation, and size evaluation of flood areas from aerial images taken with drones. First, an improved Green algorithm is designed to identify the boundaries of the flood area by integrating four sub-affected regions. Moreover, in the enclosed area, the Zigzag-Dubins trajectory planning algorithm is also designed to create a feasible detected flying trajectory so that the entire affected regions can be properly reconnoitered and surveilled. Finally, a real-time case study is implemented in Moama, Australia to demonstrate the effectiveness of the designed approaches and the flood disaster area.