Research on High-Precision SAR Image Stitching Method Based on Feature Matching and Radiometric Consistency
Acta Interdisciplinary Science · 2026
DOI: 10.48014/ais.20260705001 ↗Kurz gefasst
Fachartikel in der Zeitschrift Acta Interdisciplinary Science, 2026. Verfasst von Leiming FU.
Abstract (Originalsprache)
The limited coverage of a single SAR scene often necessitates stitching multiple overlapping images into a wide-area mosaic for large-scale applications. Feature matching and radiometric consistency are two sequential yet complementary core issues in SAR stitching: the former determines geometric accuracy, the latter determines visual naturalness, and both require joint optimization to obtain high-quality stitching results. To address speckle noise, local geometric distortion, and radiometric inconsistency, this paper designs an interactive stitching method consisting of five modules: filtering for noise suppression, feature matching combined with a deep learning network to remove mismatches, geometric correction via global or local homography, radiometric correction via histogram matching or filtering, and image fusion through direct blending or optimal seamlines with pyramid-based fusion. Experiments were conducted using Sentinel- 1 data with simulated deformations. Through comparison among seven configurations, the default configuration achieved the best results: the registration accuracy reached 1. 48 pixels, the overlapping-area MSE was 899. 68, and the effective pixel ratio was 97. 21%. The results demonstrate that the proposed method can effectively solve the core challenges of SAR image stitching and generate seamless panoramic imagery.
Zitieren
APA
FU, L. (2026). Research on High-Precision SAR Image Stitching Method Based on Feature Matching and Radiometric Consistency. Acta Interdisciplinary Science. https://doi.org/10.48014/ais.20260705001
Vancouver
FU L. Research on High-Precision SAR Image Stitching Method Based on Feature Matching and Radiometric Consistency. Acta Interdisciplinary Science. 2026. doi:10.48014/ais.20260705001
BibTeX
@article{FU2026,
author = {FU, Leiming},
title = {Research on High-Precision SAR Image Stitching Method Based on Feature Matching and Radiometric Consistency},
journal = {Acta Interdisciplinary Science},
year = {2026},
doi = {10.48014/ais.20260705001}
}
EUVIMED