Preliminary Application of High-resolution Satellite Image in City Building Classification

With the development of remote sensing technique in recent years, high resolution space image shows application potential in many fields, especially in field of city quake calamity evaluation. In this paper, Two IKONOS images of Beijing area are used to classify building types of the city. One is 1m resolution Pan image, the other is 4m resolution and multi spectrum one. After fusion processing, 1m resolution and multi spectrum image is obtained as the basic data for building classification. In order to get best results, some non spectrum information was selected to join the classification process, such as building altitude information which was obtained from the shadow of building and texture. Research shows that different characteristic combination image modes have different classified results. By comparison of four different combination modes, one can see that result, obtained by fusion image, building altitude and texture has best precision, its high building classification precision reaches 80%, and middle and low building classification precisions are all exceed the others results.