Comparison of Segmentation Algorithm for Urban Tree Delineation using Aerial Imagery

Abstract

The accurate tree delineation in urban environment is important for the efficient updating of urban planning, management as well as in the study of Urban Heat Island (UHI). In this study, trees in a selected urban area in the Terendak, Melaka are detected from aerial images by using multiresolution and feature extraction segmentation algorithm. Both segmentation results are classified using rule-based classification technique. The derived results demonstrate the ability of both object-based algorithms in delineating the tree crown in urban environment from aerial images. However, multiresolution segmentation with rule-based image classification produced a better result as compared to feature extraction’s result with 86% classification accuracy.