Object-based Vegetation Type Mapping from an Orthorectified Multispectral IKONOS Image using Ancillary Information

Traditional pixel-based image classification approaches have some limitations with the use of very high spatial resolution imagery. In recent years, object-based image analysis (OBIA) approaches has emerged with an attempt to overcome those limitations inherited to the conventional pixel-based approaches. When using OBIA approach, it is known that the quality of segmentation directly affect classification results. In this study, object-based vegetation type classifications for a steep mountain area were conducted from a multispectral IKONOS image by using spectral as well as topographic information such as elevation, aspect, and slope. In addition, another ancillary information, i.e., stream GIS data, was incorporated into image segmentation procedure. This study demonstrated that OBIA with topographic variables produced higher classification results than OBIA with only spectral information by 4.2 % and 0.04 for overall accuracy and Kappa coefficient, respectively. However, the most improved classification accuracies were acquired by using Euclidean distance as well as spectral and topographic information. In this approach, the highest classification accuracies were obtained at a scale of 48 with an overall accuracy of 76.6 % and a Kappa of 0.57. In addition, a final classification result from the scale was the most agreeable to manual interpretation. In future study, we plan to conduct a study associated with topographic correction on the multispectral IKONOS image by using a lidar-derived digital elevation model to remove spectral variation caused by terrain in the mountainous area. * Corresponding author.