Detecting Multiple Surface Targets Using a Water Surface Model

Most research topics on target detection or tracking using unmanned aerial vehicles work with ground targets. Water surface targets, on the other hand, present a new set of challenges including scene changes by wakes, sunlight reflections, and combination of the two. In this paper, we propose a method to develop water surface models considering water surface characteristics such as sunlight reflection, wakes, and water texture. Using the water surface models, we efficiently detect multiple targets on water surfaces with reduced impact of sunlight and cloud reflections.