Land use is basic information in regional and rural studies, and remote sensing (RS) is a useful tool for understanding land use and land cover (LULC). High resolution satellite images (HRSI) such as IKONOS and QuickBird have been used in LULC studies for about a decade, and they are now popular among RS professionals and nonprofessionals alike. However, classification methods are not standardized for HRSI, whereas supervised/unsupervised classification is commonly applied for middle-resolution satellite images such as Landsat.
In this study, the object-oriented classification method for HRSI is discussed in terms of LULC studies. This method has been applied in many scientific studies in the past few years, and it comes equipped with some RS software packages such as Definiens. However, the procedure to make an LULC map from HRSI has yet to be formulated and classification accuracies depend on the operator’s skills. The most significant parameter in this method is the scale parameter (SP), which determines the size of the image object. In this study, by changing SPs to the IKONOS image, it was found that the size of the homogeneous image object is influenced by the land cover type; for example, a paddy field has a larger homogeneous object size than land cover types such as residential areas. The result suggests that object-oriented land cover classification methods can be helpful for RS nonprofessionals to classify HRSIs, and the approach provides land use characteristics in the study area by understanding the land cover objects.