Object-Oriented Land Cover Classification Based on Two Satellite Images Obtained in One Dry Season in Cambodia
Some regions of the Mekong River basin still have considerable forest resources, but the pressure for exploiting these resources is very high. Changes of forest cover may strongly affect the water circulation of this region. Therefore, it is important to monitor changes of land cover of this region. In this study, we tested an objectoriented classification method to create a land cover classification map in Cambodia. A commercial object-oriented image analysis software package (eCognition) was used in this analysis. In an object-oriented classification method, the success of classification depends largely on the result of image segmentation. In this study, we overcame the difficulty in image segmentation by combining temporal images acquired in the early and late dry season. The overall accuracy was 0.70, and the Khat statistics value was 0.60. Although the accuracy was moderate, the discrimination between evergreen and deciduous forest types was good. However the mixed or the degraded land cover types were still hard to distinguish from each other. Using images taken in different phenological stages made it possible to both segment the images accurately and classify objects appropriately in an object-oriented classification process.
KeywordsLand Cover Image Segmentation Land Cover Category Shrub Land Japan International Cooperation Agency
Unable to display preview. Download preview PDF.
- Congalton RG (1991) A review of assessing the accuracy of classifications of remotely sensed data. Remote Sens Environ 37(1):35–46. Defi nens Imaging GmbH, Munich, GermanyGoogle Scholar
- Definiens Imaging (2004) eCognition version 4 object oriented image analysis user guide.Google Scholar
- FAO (2005) Global forest resources assessment 2005. FAO Forestry Paper 147. FAO, RomeGoogle Scholar
- Ministry of Public Works and Transportation, Kingdom of Cambodia and Japan International Cooperation Agency, Japan (2003) Meta-database-Cambodia Reconnaissance Survey Digital Data JICA, Tokyo, JapanGoogle Scholar
- National Office of Forest Inventory and Planning (NOFIP) (1992) Forest cover and land use in Lao PDR. Final report on the nationwide reconnaissance survey. Report no. 5. NOFIP, Vientiane, LaosGoogle Scholar
- Schriever JR, Congalton RG (1995) Evaluating seasonal variability as an aid to cover-type mapping from Landsat Thematic Mapper data in the Northeast. Photogramm Eng Remote Sens 61:321–327Google Scholar
- Wolter PT, Mlandenoff DJ, Host GE, Crow TR (1995) Improved forest classification in the Northern Lake States using multi-temporal Landsat imagery. Photogramm Eng Remote Sens 61:1129–1143Google Scholar
- Yijun C, Hussin YA (2003) Object-oriented classifier for detection tropical deforestation using LANDSAT ETM+ in Berau, East Kalimant, Indonesia. Map Asia Conference 2003 Kuala Lumpur, MalaysiaGoogle Scholar