INTEGRATION OF TEMPLATE MATCHING AND SVM TECHNIQUE FOR COCONUT TREE DETECTION

Hits: 128
Research areas:
Year:
2016
Type of Publication:
In Proceedings
Keywords:
Coconut Detection, LiDAR, SVM, Template Matching, Feature Extraction
Authors:
Bernales, Alma Mae J.; Samonte, Cristina O.; Antolihao, Julie Ann F.; Silapan, Judith R.; Edullantes, Brisneve; Pada, Ariadne Victoria S.; dela Serna, Alexis Marie L.
Abstract:
The research on feature extraction of typical objects like trees and buildings has intensified. Previous studies have demonstrated that the use of LiDAR data is very effective, especially for detailed land cover mapping. In this study, LiDAR derivatives such as Canopy Height Model (CHM), Digital Surface Model (DSM), Digital Terrain Model (DTM), LiDAR Intensity, number of returns and DSM ruggedness measure are used in image analysis in order to extract meaningful features. Detection algorithm integrating template matching and support vector machines (SVM) are then applied to extract coconut trees in the selected study area. Comparative experimental results show that this algorithm is able to detect coconut trees more effectively. The methodology of this study is very useful for the rapid assessment, monitoring and sustainability of coconut trees in the Philippines, which is the second biggest producer of coconut; but at the same time, the third most exposed countries to natural disasters.
Full text: Ab 0509.pdf [Bibtex]
You are here: Home ACRS ACRS Overview Proceedings INTEGRATION OF TEMPLATE MATCHING AND SVM TECHNIQUE FOR COCONUT TREE DETECTION