An Object-oriented Method for Extracting City Information based on High Spatial Resolution Remote Sensing Images

Abstract

The urban areas have rich spectral characteristics in high spatial resolution remote sensing images. Object-oriented technology has an extensive application in thematic information extraction and image classification fields. In terms of the extraction characteristics of object-oriented technology, this paper propose a new object-oriented method for the high spatial resolution remote sensing images classification, including spectral statistics, image segmentation, feature space optimization and fuzzy classification, and so on. Taking the SPOT5 image and Google Earth image as an example, the classification result was tested. The experiments results show that the object-oriented method has the advantage of high precision, rarely exists mistakenly and good quality of classification image.