Comparison of tree canopy extraction using Object Based Image Analysis and Deep Learning Technique in UAV images

Abstract

Unmanned aerial vehicle captured images is a new method of acquiring very high resolution data. Present days there is a huge need of precise land cover data in various fields, example land use, and natural resource conservation. For extraction of information from imagery various algorithms mostly object based has been used. Using Object Base Image Analysis methods usually achieve good results for segmentation of tree canopy in high spatial resolution aerial images. In this paper we are providing insights of segmentation algorithms used for extraction of tree canopy from very high resolution imagery. We provide detailed comparison of output generated from earlier object based tree canopy extraction algorithm with upcoming deep learning based new techniques.