Abstract
Sustainability of date palm trees based economy envisages regular monitoring of spatio-temporal variations of these trees over time. But, field based counting is a very old style of managing trees which is labour intensive, costly and time consuming while modern geo-spatial technologies provides a fast, cheaper and accurate mechanism. In this regard, the current study attempts to formulate a repeatable and robust approach utilising high spatial resolution remote sensing data integrated with image processing technique and GIS functionalities in automatically extracting palm trees. The study was carried out over Al-Ahssa region in Saudi Arabia. Difficulties posed by dark and bright canopy situation is resolved in this study based on appropriate kernel (based on multi-scale analysis) along with focal minimum and focal maximum functions, and individual tree locations were identified correctly. Supervised classification technique was utilised to differentiate trees from non-tree areas. The approach has successfully extracted date palm tree locations with an overall accuracy of 93% with an omission error of 5% and commission error of 4%.
Original language | English |
---|---|
Pages (from-to) | 579-588 |
Number of pages | 10 |
Journal | Spatial Information Research |
Volume | 28 |
Issue number | 5 |
DOIs | |
Publication status | Published - Oct 1 2020 |
Keywords
- Al-Ahssa
- Date palm trees
- High spatial resolution
- Saudi Arabia
- Tree counting
ASJC Scopus subject areas
- Computers in Earth Sciences
- Geography, Planning and Development
- Artificial Intelligence
- Computer Science Applications