Creation of DEMs Using Image Overlapping Techniques

 The creation of Digital Elevation Models (DEMs) using image overlapping techniques is rooted in the fundamental principles of stereoscopy and photogrammetry, where the spatial difference between two or more overlapping images is used to calculate elevation. This method is one of the most widely used and intuitive approaches to generating 3D surface data from 2D satellite or aerial imagery. The core idea is based on parallax—the apparent shift in the position of a surface feature when viewed from two different angles. When an area on the Earth’s surface is imaged multiple times from different positions, the displacement of features across the image pairs can be measured. These measurements are then converted into height values using geometric triangulation methods, forming a structured grid where each pixel represents a precise elevation point.

In practice, stereo pairs are generated by capturing two high-resolution images of the same location from different vantage points—often from two separate satellite passes or by a dual-camera system mounted on the same satellite (such as Cartosat-1 or SPOT). The processing software identifies matching features across the two images and computes depth by analyzing the horizontal disparity between them. Each match corresponds to a 3D point in space, and when this is done across the entire image, it results in a dense point cloud that represents the terrain. These point clouds are then interpolated to form continuous raster grids—the final DEMs.

The quality of DEMs created through overlapping image techniques depends on several factors, including the resolution of the input images, accuracy of the satellite's positioning and orientation data, the viewing angle (or baseline distance) between the image pairs, and the surface texture of the terrain. Rugged or textured surfaces like mountains and forests tend to produce more accurate DEMs due to the ease of feature matching. Conversely, flat or homogeneous areas such as deserts, snowfields, or water bodies often result in poor correlation, leading to data voids or inaccuracies.

Moreover, the use of multi-angle stereo triplets (three images taken at different angles) can significantly improve DEM accuracy by increasing redundancy and reducing the chances of mismatches. Advanced photogrammetric software automates much of this process but still requires human oversight to ensure proper image alignment, remove mismatches, and filter noisy data points. The final DEM undergoes post-processing steps such as gap filling, smoothing, and validation against ground control points (GCPs) or existing elevation datasets.

Overlapping image techniques are commonly used in aerial photogrammetry, but they are also essential in satellite missions such as ASTER GDEM and Cartosat DEM production, where image-based DEMs provide extensive elevation coverage. This approach is especially valuable for projects in inaccessible or remote regions where LIDAR or ground surveys are not feasible. The resulting DEMs are critical inputs for slope analysis, hydrological modeling, visibility analysis, and even land cover classification in GIS workflows. Complete Remote Sensing and GIS - ArcGIS – ERDAS course, where students work hands-on with real elevation datasets and learn how to visualize, process, and apply DEMs in terrain-based analysis projects.