Abstract

Despite wider applications for Unmanned Aerial Vehicle (UAV) in aquatic remote sensing, frequent sun glint in UAV acquisition often results in significant data gaps. Much research exists in the development of sun glint correction methods for airborne and satellite imagery to generate accurate coral habitat maps. Conversely, little is known about an appropriate glint correction method that can also be considered as data gap in UAV. This study compared glint correction methods for filling data gaps in UAV imagery acquired from the coral-dominated Pulau Bidong island in Peninsular Malaysia. This study proposed a simple seed pixel region growing technique that can be used in glint detection and mask development. It introduces the Theil-Sen regression glint correction (TSGC) for glint correction in UAV imagery and to achieve coral composition maps with thematic details, useful for sustainable coastal management. TSGC achieved a 25.6% greater coral classification accuracy compared to the uncorrected images.

摘要

尽管无人飞行器（UAV）在水生遥感中有更广泛的应用，但在无人飞行器获取中频繁出现太阳眩光通常会导致巨大的数据空白。在航空和卫星图像的太阳闪烁校正方法的开发中，已有许多研究工作，以产生准确的珊瑚栖息地图。相反，对于适当的闪烁校正方法知之甚少，该方法也可以视为无人机中的数据间隙。这项研究比较了闪烁校正方法来填补从马来西亚半岛珊瑚礁为主的碧东岛获取的无人机图像中的数据空白。这项研究提出了一种简单的种子像素区域生长技术，可用于闪烁检测和掩模开发。它介绍了Theil-Sen回归闪烁校正（TSGC），用于在无人机图像中进行闪烁校正，并获得具有主题细节的珊瑚组成图，这对可持续的沿海管理非常有用。与未经校正的图像相比，TSGC的珊瑚分类精度提高了25.6％。