Emrah Onat; Yakup Özkazanç · 2026 · Turkish Journal of Remote Sensing
Paper
This study investigates the influence of various interferogram filtering techniques on the accuracy of Digital Elevation Models (DEMs) generated through Synthetic Aperture Radar Interferometry (InSAR). While most studies in the literature evaluate filter performance solely based on interfero-gram-domain metrics, this work extends the analysis to assess how denoising affects the accuracy of the final DEM product. Both conventional filters—such as mean, median, pivoting median, Lee, Frost, Modified Frost, Kuan, Kuwahara, and Baran—and advanced algorithms including NL-InSAR, PPB, MuLoG, BM3D, and InSAR-BM3D are systematically compared. Simulations are performed on both synthetic and real SAR datasets to ensure reliability and generalizability. The filters are evalu-ated using metrics such as Peak Signal-to-Noise Ratio (PSNR), Edge Preserving Index (EPI), Sum of Phase Differences (SPD), and number of residues, while DEM accuracy is quantified through Root Mean Square Error (RMSE). The results aim to identify the most effective denoising strategies for enhancing DEM generation and to provide guidance for future InSAR-based terrain mapping ap-plications.
Analysis
This study evaluates the impact of various interferogram filtering techniques on the accuracy of InSAR-derived Digital Elevation Models (DEMs).
Source record