Land Use/cover Change Detection with Multi-source Data

It is of great importance to obtain the earth surface property timely and effectively,which can help us to know the relationship between human and nature phenomena and also for decision-making.Pixel-based and classification based remote sensed data are the two normal methods during the traditional land use/cover change detection,which make use of the single-source spectral to extract the changed land use/cover information,while texture and other spatial information are neglected.In this research,spectral and texture information basing on the Change Vector Analysis and Support Vector Machine method are incorporated to extract the land use/cover information.The land use/cover information are extracted with the method above in Haidian district,Beijing,supported with the two-temporal TM image in 1997 and 2004,the overall accuracy and Kappa are 93.1% and 0.862 respectively,better than double-windows flexible pace searching CVA method.Whose overall accuracy and Kappa are 90.2% and 0.804 respectively,showing that the method in this paper can extract the changed information effectively.On the other hand,this method can over come the difficulty in searching the threshold which has to be engaged in the CVA method.