Researchers: Sören Dittmer, Jonathan Williams, Carola-Bibiane Schönlieb, Tom Swinfield, David A. Coomes, Juheon Lee, Xiaohao Cai

We apply graph cut approaches to detect, segment and analyse the structure of individual trees in remote sensing data over forested landscapes. Our approach takes account of knowledge of typical tree geometry and local variations in 3D structure of the data to separate individual tree crowns (ITCs). Our focus has been on working with Light Detection and Ranging (LiDAR) data, but through previous work on co-registration of other remote sensing data type (hyperspectral and RGB imagery) we are able to refine tree detections based on underlying spectral properties of the trees.

This project is part of the one of our collaborative projects: INTEGRAL