DEM densification using SFS with single multi-spectral satellite image

摘要

As for the shortcoming that traditional interpolation methods often cause the over-smooth problem, or couldn't fully take the variety of the terrain detail into account, this paper proposed a DEM densification method by using shape from shading (SFS) based on spectral information from single highly spatial resolution satellite image. In accordance with the idea of introducing SFS into DEM interpolation, a method is put forward, which is under the condition of the unknown light source in spatially heterogeneous area. Surface relative shape was reconstructed at first, and the second order edge-oriented image interpolation method was applied to generate a high-resolution DEM grid. Spectral information of the unknown points was used to reveal the actual surface reflection properties, and land surface could be accurately modeled compared traditional SFS method, which use a constant reflectance in the whole region. Experiments proved that the algorithm is very effective for the sparse grid DEM interpolation and offer a new way for DEM densification.