DEM extraction based on SFM using remote sensing images

Abstract

Digital Elevation Model (DEM) extraction from unmanned aerial vehicles is one of focal points in remote sensing image processing. This paper built point clouds based on Structure From Motion (SFM) techniques firstly, then excluded the outliers from the point clouds, and finally generated the regular grid DEM through different interpolation methods. At the same time, the paper proposed an improved inverse distance weighted interpolation method based on the exponent p of weighting function. As seen in the experimental results evaluated by checkpoint method, the improved inverse distance weighted interpolation got the higher accuracy than inverse distance weighted interpolation. The followed was nearest neighbor interpolation. The moving average interpolation performed the worst. The results indicated that the improved interpolation method was an effective way to extract DEM using photogrammetry techniques.