Skip to Main Content
In recent years the land-use/cover change is one of the important points of the global climate change. The remote sensing techniques provide strong support to the study on the wide land-use/cover change. Because of the common existing of same object having different spectral character and different object having same spectral character, the classification accuracy of traditional pixel-based has not yet satisfied the need of the monitor of the land-use/cover change. The newly object-oriented method opens a new path for the remote sensing classification. The biggest contribution is that the new method makes the theory of the abstraction of the remote sensing characteristics be perfect. Originally, it is difficult to extract the relationship of shape, location and space, now the object-oriented method makes it be possible in the condition with remote sensing data of higher spatial resolution. In this paper, these two methods ware carried out for TM data based on spectral, texture, and shape features, and the classification accuracy was compared and analyzed with that of man-expert visual interpretation. The results show that (1) TM data are also fit to the method of object-oriented classification. (2) The accuracy of object-oriented method is higher than that of pixel-based method, and the classification result has less pepper-and-salt noise, omitting trivial classification past-processing. (3) The optimal texture features group by the two methods is very similar in the smaller calculation window. (4) The classification effects with shape feature in TM data source are not outstanding.