Highly accurate geometric correction is fundamental to the application of satellite images. In highly accurate geo- metric correction of remote sensing images for precise match to the base maps

the difficulties in capturing ground control points （GCPs）

insufficient number of GCPs and low accuracy of the geo-referenced images remain to be frequent issues. In search for solutions to these issues

this paper proposes a new method for the geometric correction of satellite images for the mountain ar- eas based on the terrain feature lines extracted from the Digital Elevation Model （DEM）. This method uses DEM as the base maps without any geometric distortion for the geometric correction of satellite images. This paper presents the theories and method to extract terrain feature lines

such as gorges

mountain ridges

peaks and concaves and describes the procedures for the geometric correction for satellite images of mountain areas based on the terrain feature lines. It further discusses the problems in extracting terrain feature lines and capturing of GCPs

as well as the solutions to these issues. The outcome from the experiments shows the number of terrain feature lines of mountains areas are a great many times of those in the conventional map layers of hydrological systems and road networks

and that terrain feature lines extracted from DEM present high stability and reliability and can be applied in the highly accurate geometric correction for satellite images of mountain areas. The results of statistical test show the average geometric error is reduced to the size of one pixel when this method is used in geometric correction.