Abstract: A classification method based on topographical area segmentation is proposed.In the method,the disturbed areas of objects are divied by topographical features and the area-based classification with MLC is applied in the image.Meanwhile,the topographical comection with DEM data is adopted in order to decreasing the scattev of spectrum in the same or the similar objects caused by irregular terrain.The experiment by a landsat image in west of Hubei province proved that the results of area-based classification have obvious improved than that of the traditional spectral classification.