Extracting water information from QuickBird images based on object-oriented method

The traditional classification and extraction of information about remote sensing is based on spectral information of remote sensing images with man-made interpretation method. Because of the existence of the same objects with different spectrums and different things with same spectrums and the subjectivity of the man-made interpretation, the classification accuracy of this method is low and inefficient. However, the classification technology based on objects adopts the method of multi-scale segmentation to produce the image objects, using the spectral characteristics of remote sensing images. Then, based on the geometric features and topological relations of images objects, the rules of classification are established which are used to extract the information of the images. Studies have shown that this method has better results. In this paper, extraction of water baby information in XiaoTangShan area, as an example, confirms the truth of this method.