Mathematical morphology has already become the important research field of computer image processing

and it has been applied effectively in the remote sensing image analysis. This paper introduces mathematical morphology into the research of building DEM

provides a algorithm that those homologous points can be auto recognized from stereo images by using image matching and the 3D coordinates of ground points are extracted

then the DEM can be built by using morphologic transformation. The research injects new vigor into the arithmetic of auto building DEM. This paper discusses the method of using orthogonal wavelet to decompose and reconstruct a image?feature extraction and matching combining image and feature . The algorithm for auto building Thiessen polygon

Delaunay triangular net (TIN) and regular net (GRID) of DEM is presented with the discrete ground points from stereo matching . Experiments prove that this arithmetic has the virtue of simple data construction

high speed

efficient and high accurate.