Abstract
The interpolation algorithms for slave SAR image resampling were studied and com-pared. The algorithms nearest neighbor, bilinear interpolation, cubic convolution and advanced methods such as Raised Cosine kernel, Knab interpolation kernel and truncated sinc kernel were tested on three pairs of ERS tandem data.
The coregistration performance was evaluated by means of coherence and the sum of phase differences. An influence of a used resampling method on the accuracy of the final DEM was also examined.
The national DEM of the Czech Republic, ZA-BAGED, and a newly created DEM from airborne laser scanning were utilized for InSAR DEM validation. No evidence showed that the computationally intensive algorithms produce better re-sults than the cubic convolution method.