Object identification using multi-resolution satellite sensor images

During the past decade, spatial topographic mapping from space using multi-resolution satellite sensor data instead of conventional data collection techniques widely gained popularity. The recognition and identification of spatial objects on satellite images is rarely being a problematic issue because of the pixelization. The aim of this study is to bring an empirical assessment for the identification of spatial objects using multi-resolution satellite sensor images. This study also includes statements about the terms recognition and identification in means of remote sensing and cartographic techniques. Satellite images of different resolutions like LANDSAT-5, SPOT-4, SPOT-5 and IKONOS and the 1:5000 scaled topographic maps are used. The maps and images cover the provinces around the Bosphorus, Istanbul. 16 sample objects are selected in the study area, which are known and common textures on all images and maps. The area is the size of interest for polygonal objects and the width is for the linear objects. The views of these samples in different images have been compared mutually for determination of recognition and identification limits with human sight. Spatial filtering and enhancement techniques are realized for the raster images. Their effects to the visual identification of objects are also taken into account by the analysis. The results are interpreted due to the different map scales.