Two-Step Approach toward Alignment of Spatiotemporal Wide-Area Unmanned Aerial Vehicle Imageries

Recently, analysis and decision-making based on spatiotemporal unmanned aerial vehicle (UAV) high-resolution imagery are gaining significant attention in smart agriculture. Constructing a dataset requires multiple UAV image mosaics taken at different times. Because the weather or flight trajectory is subject to change when images taken, typically unaligned. This paper proposes two-step approach, composed of global local alignments, for alignment two wide-area high resolution. The first step, alignment, finds projection matrix that initially maps keypoints source mosaic onto matched counterparts target mosaic. next refines result alignment. proposed method splits input into patches applies individual transformations each patch enhance remaining misalignments level. Such independent alignments may new artifacts boundaries. uses simple yet effective technique suppress those without harming benefit Extensive experiments validate by using several datasets highland fields plains South Korea. Compared with recent work, improves accuracy up 13.21% over datasets.