The use of object based classification with nDSM to increase the accuracy of building detection

Abstract

Satellite technologies are used in various disciplines with the improvements in resolutions of sensors. Although spatial resolution of remotely sensed images has increased, automatic information extraction from these images could not reach the desired accuracy. Different studies have been conducted for eliminating the automatic information extraction errors and producing highly accurate spatial information from these images. In this study, building detection analysis was performed with object oriented classification of high resolution satellite images. Results of the object based classification were compared with the integration of normalized digital surface model (nDSM) and NDVI to the object based classification and the effects of these additional data on the classification accuracy were evaluated.