Study on a kind of land-use change detection based on patch objects of remote sensing image

Abstract

Object-oriented change detection is more practical than per-pixel change detection when we survey change information of land-use using remote sensing image. A kind of land-use change detection based on patch objects of remote sensing image is studied in this paper. The patch objects of image come from the polygons of historical vector land-use map. The method calculates the weighted correlation coefficient of each patch object in order to identify whether the patch object is changed. Two different time-series SPOT images in a suburb of Shanghai are used to verify the method. The result shows that the method can detect the changes effectively and raises automation level of land-use updating investigation.