Accurate Detection of Occluded Wildfire Boundary

Abstract Inaccurate localization of wildfire boundary will adversely affect the efficiency fire fighting, especially in case limited water for aerial dumping. This paper formulates detection occluded as a shape completion problem and introduces new solution. Different from traditional algorithms which only consider contour geometry, proposed algorithm takes partially visible surface geometry into consideration, generates extrapolated feature points to detect invisible section. The solution involves two processes: firstly, surfaces above segments are ground level with Hurwitz–Radon operator based obtain boundary; secondly, interpolating extra control Bezier curve reduce uncertainty improve accuracy detection. To demonstrate effectiveness algorithm, series numerical simulations were performed. Simulated results show that this study can conforms truth boundary.