BIOGRAPHY

Hamdan Omar
Hamdan Omar is a Research Officer at the Geoinformation Programme, Forestry and Environment Division, FRIM. He is also pursuing his PhD in the field of forest engineering and operations at Universiti Putra Malaysia (UPM). His work is mostly concentrated on the use of remotely sensed data in forestry applications.

Abstract
Performances of UAV and WorldView-2 Images for Individual Canopy Delineation in Tropical Forest
Recent developments in high spatial resolution remote sensing have created a wide array of potential new forestry applications. High spatial resolution imagery allows a tree-scale of analysis, in which individual trees and their attributes are the focus of interest. Currently, there are many airborne and spaceborne sensors that offer high spatial resolution images. Along with the advancements in remote sensing technology, various methods have been developed to delineate individual tree canopy in different type of vegetations and ecosystems. In this study, aerial photos acquired from unmanned aerial vehicle (UAV) and images from WorldView-2 (panchromatic band) satellites were used to delineate individual tree canopies. These images have spatial resolution of 0.1 and 0.5 m, respectively and both were acquired in year 2013. The information was used to quantify how many trees stand in the study area. The campus of Forest Research Institute Malaysia (FRIM) was selected as the study area. The study covered the entire 540 ha of FRIM campus, out of which 420 ha is covered with forest comprises natural and planted forests. Object-based or segmentation approach was employed to delineate the canopies. A delineated canopy was assumed to corresponding to an emergence tree. Results generated from both images were compared and the canopies counted were validated via ground thruthing data. The study indicated that the data from UAV produced more accurate results (i.e. ~90%) than the WorldView-2 images (i.e. ~78%). However, these accuracies increased in planted forest that has uniform and homogenous canopy structure. In some cases, the UAV data has overestimate the trees due to its spatial capability that counted more canopies for a single tree. The performances of high spatial resolution imageries, especially of lower than 0.5 m were examined and both indicated improvements in precision forestry applications. Further exploration on methods used to delineate individual canopy, especially in complex tropical forest need to be done to obtain more accurate results and to make sure that the technology is fully exploited in the future.
Malaysia Geospatial Forum 2014 Secretariat
D-5-13A, Block D, Capital 4, Oasis Square, No. 2
Jalan PJU 1A/7A, Ara Damansara
47301 Petaling Jaya, Selangor
Tel : +6 03 7859 8411 Fax : +6 03 7859 8412 Email : info@malaysiageospatialforum.org