To improve the quality and efficiency of digital elevation model (DEM) construction in complex terrain, this study proposes a multi-source DEM acquisition and fusion method that integrates high-resolution optical imagery and interferometric synthetic aperture radar (SAR) imagery. Using an unmanned aerial vehicle and satellite remote sensing system as a platform, this method constructs a multi-view data acquisition chain to generate optical imagery DEM and interferometric SAR-DEM, respectively. By introducing a point cloud classification algorithm based on texture and structural features and a regional adaptive weight estimation model, achieving weighted fusion of multi-source elevation data. The fusion process employs error constraints and edge control strategies to address typical challenges such as terrain occlusion, data holes, and elevation jumps. Experimental validation in representative landforms, including forests, glaciers, deserts, cities, and water bodies, demonstrates that this method achieves excellent elevation recovery accuracy and boundary continuity, adapting to the three dimensions modeling needs of various landform types. This research provides stable and reliable technical support for applications such as high-resolution terrain mapping, landform evolution monitoring, and disaster early warning, and is of great significance for advancing the automation and intelligence of remote sensing mapping.