TREETOP DETECTION FOR A RIVERINE BAMBOO FOREST USING UAV IMAGES

Abstract

 Treetop detection was conducted at a dense bamboo (Phyllostachys nigra var. henonis) forest using local maximum filtering (LMF) with a circular window under a high-resolution digital surface model (DSM) via UAV images. Optimal window size (WS) was evaluated based on applications in some WS cases. Treetop detection results showed that the highest F-value was obtained in 0.5-m WS, with a detection accuracy that was relatively higher than that of a previous study, although in a denser bamboo forest. Thus, high-resolution DSM and an appropriate WS are important factors for accurate treetop detection by LMF.