After the high resolution images became commercially available, 3D surface models generated from space-born stereo images turned into an attractive alternative for applications such as telecommunication planning, disaster monitoring and urban planning. The accuracy of the 3D models of the earth surface depends on the accuracy of corresponding points located in both images. Area-based automatic image matching combined with a region-growing technique are able to provide a dense and accurate grid of corresponding points. However the region-growing process may stop at image patches where the effect of a sudden change in the x-parallax is important. In such a case new seed points must be provided, usually by human operator. From the additional seed points the region-growing procedure may continue. Depending upon the type of image and the 3D-structure of the mapped area, the human intervention may be considerable. A fully automatic alternative that combines the SIFT (Scale Invariant Feature Transform), least square matching and region-growing technique is proposed in this work. Experiments conducted on stereo pairs of Ikonos and aerial images covering different terrain types have shown the effectiveness of the proposed method especially in locations with abrupt height changes, such as façades of high buildings.