Tropical Agrarian Landscape Classification using high-resolution GeoEYE data and segmentationbased approach

We examine the use of high spatial resolution ‘GeoEYE’ imagery for land use classification in a tropical landscape. Image objects (I-Os) derived from features identification provide a basis for segmentation process and the Geographic Object Based Image Analysis (GEOBIA) framework. eCognition software with I-Os as classification unit and maximum likelihood algorithm facilitated the process. Supervised classification approaches (SCA) and rule set classification approach (RSCA) were used and performance and transferability of two approaches compared. Main conclusions: (a) high degree of details in GeoEYE data enables delineation of diverse land use zones, and (b) segmentation based analysis is more effective to tackle spatial intermixing.